View Recording: Milwaukee AI Summit 2026 – Governing & Securing AI at Scale

1 00:00:12,045 –> 00:00:13,246 name is Joe Steiner. 2 00:00:13,613 –> 00:00:14,881 I’m a solutions architect 3 00:00:14,881 –> 00:00:16,082 with Concurrency. 4 00:00:16,249 –> 00:00:17,250 We’re going to be talking about 5 00:00:17,250 –> 00:00:18,451 AI governance. 6 00:00:18,618 –> 00:00:22,122 How do we govern and secure AI? 7 00:00:22,122 –> 00:00:24,124 As we we kind of work 8 00:00:24,124 –> 00:00:25,458 along the journey into that. 9 00:00:25,725 –> 00:00:27,460 Each of you are probably in different places there. 10 00:00:27,460 –> 00:00:29,329 So we’re going to kind of talk about the progression there. 11 00:00:29,496 –> 00:00:30,730 It’s really the best way to approach that. 12 00:00:30,730 –> 00:00:32,165 You’re not going to solve it all at once. 13 00:00:32,632 –> 00:00:33,600 And so we’re 14 00:00:33,600 –> 00:00:35,201 going to talk about, what the best way 15 00:00:35,201 –> 00:00:36,436 to work your way through that is, 16 00:00:36,803 –> 00:00:38,238 after this, 17 00:00:38,238 –> 00:00:39,439 we’ve got what I view to be 18 00:00:39,439 –> 00:00:40,640 the highlight of the day, 19 00:00:40,774 –> 00:00:41,708 which is we have 20 00:00:41,708 –> 00:00:43,176 some panelists that are going to be coming 21 00:00:43,176 –> 00:00:44,377 from the audience 22 00:00:44,444 –> 00:00:45,545 to talk about 23 00:00:45,545 –> 00:00:46,780 their real world experiences 24 00:00:46,780 –> 00:00:48,014 with AI, what they’ve done. 25 00:00:48,281 –> 00:00:49,449 And I’m really looking forward to that. 26 00:00:49,449 –> 00:00:51,718 So I hope you’ll stick around with us for that. 27 00:00:52,352 –> 00:00:55,722 So we talk about AI. 28 00:00:55,889 –> 00:00:59,526 Just kind of backing up a bit. 29 00:00:59,592 –> 00:01:00,960 Humans have had kind of 30 00:01:00,960 –> 00:01:02,662 a weird relationship with AI 31 00:01:03,029 –> 00:01:04,631 for about 75 years. 32 00:01:05,265 –> 00:01:06,299 When you think about it, 33 00:01:06,299 –> 00:01:07,867 you know, back in the 50s, 34 00:01:07,867 –> 00:01:09,235 there are stories about 35 00:01:09,235 –> 00:01:11,404 Asimov started with this, and there have been plenty 36 00:01:11,404 –> 00:01:13,039 of other science fiction has been addressing 37 00:01:13,039 –> 00:01:14,240 this for some time. 38 00:01:14,574 –> 00:01:15,842 The concerns about what 39 00:01:15,842 –> 00:01:17,677 AI is going to do to the world, 40 00:01:17,677 –> 00:01:18,878 to our lives, to our 41 00:01:19,145 –> 00:01:21,514 our relationships with it, 42 00:01:21,514 –> 00:01:23,049 spanned a spectrum here 43 00:01:23,049 –> 00:01:25,618 between kind of those doomsday scenarios. 44 00:01:25,618 –> 00:01:27,120 You saw that with like Terminator 45 00:01:27,120 –> 00:01:28,388 and the Skynet stuff, 46 00:01:28,388 –> 00:01:31,324 and on to Utopia, where everything’s going to be, 47 00:01:31,324 –> 00:01:32,792 you know, golden and great and, 48 00:01:33,126 –> 00:01:34,494 no worries. Right. 49 00:01:35,028 –> 00:01:36,229 The truth is, 50 00:01:36,729 –> 00:01:38,231 those make for good stories. 51 00:01:38,665 –> 00:01:39,833 Make for good movies. 52 00:01:39,833 –> 00:01:41,234 But not really 53 00:01:41,234 –> 00:01:42,435 where we’re heading, right? 54 00:01:42,669 –> 00:01:43,636 Really where we’re going to be 55 00:01:43,636 –> 00:01:45,405 is it’s up to us to determine 56 00:01:45,405 –> 00:01:47,841 whether AI is detrimental 57 00:01:47,841 –> 00:01:49,042 to organizations. 58 00:01:49,109 –> 00:01:50,643 It’s kind of neutral, doesn’t do much 59 00:01:51,010 –> 00:01:52,579 or to be realized. 60 00:01:52,579 –> 00:01:54,280 The possibilities of this really having 61 00:01:54,280 –> 00:01:56,549 some exponential improvement 62 00:01:56,783 –> 00:01:57,851 in what we’re doing. 63 00:01:57,851 –> 00:01:59,185 But it’s on all of us to how we 64 00:01:59,252 –> 00:02:00,386 how we get there. 65 00:02:02,255 –> 00:02:03,523 As we’ve gone through 66 00:02:03,556 –> 00:02:05,992 history with different, technologies, 67 00:02:05,992 –> 00:02:07,193 you know, started 68 00:02:07,293 –> 00:02:08,661 that have started way back, 69 00:02:08,661 –> 00:02:09,996 but you look back at AI, 70 00:02:09,996 –> 00:02:12,398 I look at what we saw with, 71 00:02:12,398 –> 00:02:13,600 the internet. 72 00:02:13,800 –> 00:02:14,134 Right. 73 00:02:14,134 –> 00:02:15,702 It’s a similar adoption 74 00:02:15,702 –> 00:02:16,903 pattern there. 75 00:02:16,970 –> 00:02:19,105 And there are different ways to approach it. 76 00:02:19,239 –> 00:02:20,473 So the same thing with cloud. 77 00:02:20,907 –> 00:02:23,076 So some organizations 78 00:02:23,076 –> 00:02:24,277 will start with a hey, 79 00:02:24,377 –> 00:02:25,812 we’re going to avoid using this. 80 00:02:25,812 –> 00:02:26,112 Right. 81 00:02:26,112 –> 00:02:27,580 We’re going to kind of let this settle out, 82 00:02:27,580 –> 00:02:28,781 see what happens. 83 00:02:29,449 –> 00:02:30,650 The problem with that 84 00:02:30,650 –> 00:02:31,851 is that you don’t realize 85 00:02:31,851 –> 00:02:33,820 any of those benefits until you start 86 00:02:33,820 –> 00:02:34,988 using it. 87 00:02:35,488 –> 00:02:36,389 The other problem with that, 88 00:02:36,389 –> 00:02:37,590 and the bigger problem in my mind, 89 00:02:37,590 –> 00:02:38,858 is that frequently 90 00:02:38,858 –> 00:02:40,960 organizations will take an approach like, hey, 91 00:02:41,261 –> 00:02:42,762 we’re going to ban this, right? 92 00:02:42,762 –> 00:02:43,963 We’re going to wait till 93 00:02:43,963 –> 00:02:45,165 we’re all set. 94 00:02:45,398 –> 00:02:46,199 That leads to what 95 00:02:46,199 –> 00:02:47,667 I would describe as your shadow IT 96 00:02:47,767 –> 00:02:48,902 Mutiny. 97 00:02:48,902 –> 00:02:51,704 We saw this in ten years ago. 98 00:02:51,704 –> 00:02:53,173 I had a client here in Wisconsin, 99 00:02:53,173 –> 00:02:55,875 large manufacturing, company 100 00:02:55,875 –> 00:02:57,911 that was banning cloud technology. 101 00:02:58,545 –> 00:02:59,746 They’re people. 102 00:02:59,979 –> 00:03:01,848 When in response to this, said, well, 103 00:03:01,848 –> 00:03:03,783 I still need this to get my job done right. 104 00:03:03,783 –> 00:03:05,118 This is kind of ridiculous. 105 00:03:05,618 –> 00:03:07,187 So they literally would leave 106 00:03:07,187 –> 00:03:09,956 the office, take a personal computer, 107 00:03:10,256 –> 00:03:11,758 take the files from their work, 108 00:03:12,225 –> 00:03:13,693 put it on a personal computer, 109 00:03:13,927 –> 00:03:15,128 go to a coffee shop, 110 00:03:15,128 –> 00:03:17,130 literally across the street. Right. 111 00:03:17,163 –> 00:03:18,331 Some of you some of you know what I’m 112 00:03:18,331 –> 00:03:19,532 talking about, 113 00:03:19,799 –> 00:03:21,034 literally across the street, 114 00:03:21,534 –> 00:03:22,468 get their work done 115 00:03:22,468 –> 00:03:23,670 and then come back in. 116 00:03:23,803 –> 00:03:25,905 All for legitimate purposes, right? 117 00:03:25,905 –> 00:03:27,373 They weren’t doing anything nefarious. 118 00:03:27,373 –> 00:03:28,875 They’re not trying to get away with anything. 119 00:03:29,175 –> 00:03:30,376 It’s trying to get work done. 120 00:03:30,944 –> 00:03:32,145 And it got kind of ridiculous. 121 00:03:32,212 –> 00:03:33,746 Got particularly ridiculous when they started 122 00:03:33,746 –> 00:03:35,815 noticing other employees doing the same thing, 123 00:03:35,815 –> 00:03:37,083 sitting at the coffee shop with them. 124 00:03:37,483 –> 00:03:38,418 I mean, it’s just it’s 125 00:03:38,418 –> 00:03:40,019 it’s moronic, right? 126 00:03:40,787 –> 00:03:42,655 So once everyone starts 127 00:03:42,655 –> 00:03:43,856 noticing that, 128 00:03:44,157 –> 00:03:44,757 that starts 129 00:03:44,757 –> 00:03:46,426 leading to employee dissatisfaction, 130 00:03:47,193 –> 00:03:48,161 then you have people starting 131 00:03:48,161 –> 00:03:49,329 to realize, like, hey, 132 00:03:49,596 –> 00:03:51,130 this is a lot harder to work here 133 00:03:51,431 –> 00:03:53,433 than it is somewhere else, right? 134 00:03:53,466 –> 00:03:55,835 I’m going to go somewhere that’s going to let me do the job 135 00:03:55,835 –> 00:03:57,036 the right way. 136 00:03:57,203 –> 00:03:58,905 Which creates all kinds of new risks 137 00:04:00,506 –> 00:04:01,708 on the other end, 138 00:04:02,575 –> 00:04:02,842 when you 139 00:04:02,842 –> 00:04:04,844 have new technologies, you’ll have organizations 140 00:04:04,844 –> 00:04:07,146 that say, hey, this is the best thing ever. 141 00:04:07,247 –> 00:04:07,981 Don’t worry about it. 142 00:04:07,981 –> 00:04:09,349 Let’s just let’s dive 143 00:04:09,349 –> 00:04:10,550 in headfirst, right? 144 00:04:11,050 –> 00:04:12,418 And so it’s great. 145 00:04:12,452 –> 00:04:14,587 You start realizing benefits early. 146 00:04:15,255 –> 00:04:16,456 You don’t restrict 147 00:04:16,889 –> 00:04:17,890 what’s what’s happening there. 148 00:04:17,890 –> 00:04:19,659 So you can realize kind of the maximum 149 00:04:19,659 –> 00:04:20,860 level of benefits, 150 00:04:21,227 –> 00:04:22,695 not calculating for the risks. 151 00:04:23,696 –> 00:04:24,564 There are massive 152 00:04:24,564 –> 00:04:26,399 risks with doing that. Right. 153 00:04:26,432 –> 00:04:28,935 So I might I have unmanaged 154 00:04:28,935 –> 00:04:30,937 usage of these new techs. 155 00:04:31,638 –> 00:04:33,006 There’s a sorted 156 00:04:33,439 –> 00:04:34,807 security problems, 157 00:04:35,108 –> 00:04:37,577 operational issues, financial issues 158 00:04:37,577 –> 00:04:39,045 ultimately that come from that. 159 00:04:39,879 –> 00:04:40,546 And ultimately 160 00:04:40,546 –> 00:04:41,748 when you have an unmanaged 161 00:04:41,748 –> 00:04:43,116 technology in the environment 162 00:04:43,483 –> 00:04:44,817 it starts to lead to employee 163 00:04:44,817 –> 00:04:46,019 distrust of that. 164 00:04:46,419 –> 00:04:47,620 So part of the problem with 165 00:04:47,620 –> 00:04:48,988 that is that if I’ve got something 166 00:04:48,988 –> 00:04:50,189 that’s not managed, 167 00:04:50,256 –> 00:04:51,424 particularly with AI, 168 00:04:52,525 –> 00:04:53,226 I’m using these 169 00:04:53,226 –> 00:04:55,295 tools, it presenting me bad information. 170 00:04:55,628 –> 00:04:56,629 I’m making decisions 171 00:04:56,629 –> 00:04:57,830 off of bad information. 172 00:04:58,431 –> 00:04:59,632 That’s not good. 173 00:04:59,666 –> 00:05:01,701 People start to have a their opinion of 174 00:05:01,701 –> 00:05:03,970 AI soured from things like that. 175 00:05:04,304 –> 00:05:05,672 So you got to manage that too. 176 00:05:06,673 –> 00:05:07,874 The truth is, 177 00:05:08,641 –> 00:05:09,842 best way to approach it 178 00:05:10,209 –> 00:05:11,744 as it is with so many things in 179 00:05:11,744 –> 00:05:13,479 life is kind of in the middle. 180 00:05:14,213 –> 00:05:15,982 How do we establish kind of a responsible 181 00:05:15,982 –> 00:05:17,183 use model for this? 182 00:05:17,350 –> 00:05:19,018 Realize maybe not all benefits, 183 00:05:19,018 –> 00:05:20,219 but most 184 00:05:20,620 –> 00:05:22,455 we have managed usage of this. 185 00:05:22,655 –> 00:05:24,791 Try and drive towards employee satisfaction 186 00:05:24,791 –> 00:05:25,992 and there as best we can 187 00:05:26,392 –> 00:05:27,727 and minimize the risk. 188 00:05:28,161 –> 00:05:29,729 Right. To do this. 189 00:05:29,796 –> 00:05:31,064 That’s where governance comes in. 190 00:05:33,199 –> 00:05:34,567 For me, AI governance 191 00:05:34,834 –> 00:05:37,337 is about enabling responsible AI use 192 00:05:38,438 –> 00:05:39,639 and development 193 00:05:39,706 –> 00:05:41,274 by focusing on certain things. 194 00:05:41,808 –> 00:05:43,443 One is certainly the risks. 195 00:05:43,810 –> 00:05:45,745 This is not just security risks. 196 00:05:46,212 –> 00:05:47,513 These are operational risks. 197 00:05:47,547 –> 00:05:48,815 These are financial risks. 198 00:05:49,115 –> 00:05:50,817 There’s a host of things to be part of this. 199 00:05:50,983 –> 00:05:53,052 Security is a big part of this, to be sure. 200 00:05:53,486 –> 00:05:54,620 But there’s more than that. 201 00:05:55,888 –> 00:05:56,689 The other 202 00:05:56,689 –> 00:05:57,824 side thing that you’re balancing 203 00:05:57,824 –> 00:05:59,359 there is how am I balancing 204 00:05:59,359 –> 00:06:01,761 that with maximizing the business value 205 00:06:01,761 –> 00:06:02,962 that I’m trying to realize 206 00:06:03,296 –> 00:06:04,230 from, like Brian’s talk 207 00:06:04,230 –> 00:06:05,531 this morning, where he’s talking about 208 00:06:05,665 –> 00:06:07,734 creating a portfolio of of agents, 209 00:06:07,734 –> 00:06:08,935 how do I’m designing that? 210 00:06:09,202 –> 00:06:09,969 Somebody is kind 211 00:06:09,969 –> 00:06:11,270 of come forward to the vision for what 212 00:06:11,270 –> 00:06:12,472 they want out of AI. 213 00:06:12,605 –> 00:06:14,540 Governance has a responsibility to help them 214 00:06:14,540 –> 00:06:16,342 realize that in a managed, 215 00:06:16,342 –> 00:06:17,844 safe fashion. Right? 216 00:06:17,877 –> 00:06:19,178 That’s really what this is about 217 00:06:19,679 –> 00:06:20,847 ensuring usability. 218 00:06:20,847 –> 00:06:21,581 Make sure that you’re not 219 00:06:21,581 –> 00:06:23,716 locking things down so much that people can’t use it. 220 00:06:24,117 –> 00:06:25,551 Kind of like my example earlier. 221 00:06:26,786 –> 00:06:28,287 Also, how am I making sure 222 00:06:28,287 –> 00:06:29,489 that I’m maintaining trust 223 00:06:29,956 –> 00:06:31,391 so that people are using this? 224 00:06:31,424 –> 00:06:32,625 We’ve made this available 225 00:06:33,025 –> 00:06:34,560 and the the information 226 00:06:34,560 –> 00:06:35,928 provided is of a quality 227 00:06:36,362 –> 00:06:38,564 that people can trust that they’re not 228 00:06:38,564 –> 00:06:40,833 having to constantly second guess everything. 229 00:06:42,068 –> 00:06:43,436 Key areas for doing this 230 00:06:43,436 –> 00:06:45,705 involve setting the right policy and strategy 231 00:06:46,139 –> 00:06:47,340 for the organization, 232 00:06:48,541 –> 00:06:50,243 for putting in the right education 233 00:06:50,243 –> 00:06:52,678 and enablement efforts in there, enabling the people. 234 00:06:52,712 –> 00:06:54,280 This is a very much a people 235 00:06:54,280 –> 00:06:56,115 first technology in so many ways. 236 00:06:57,517 –> 00:06:59,152 How do I put the right protections 237 00:06:59,185 –> 00:07:00,386 monitoring in there? 238 00:07:01,687 –> 00:07:01,988 And then 239 00:07:01,988 –> 00:07:04,390 how do I ensure that I’ve got any kind of ongoing 240 00:07:04,424 –> 00:07:06,225 testing and development that happens here too? 241 00:07:06,659 –> 00:07:07,660 It’s not just a set 242 00:07:07,660 –> 00:07:08,828 at once and forget it. 243 00:07:09,095 –> 00:07:09,929 You need to continue 244 00:07:09,929 –> 00:07:11,164 to manage this over time. 245 00:07:11,330 –> 00:07:13,065 Now when I look at I use 246 00:07:13,599 –> 00:07:15,301 there’s kind of a spectrum of that, 247 00:07:15,601 –> 00:07:16,536 right? 248 00:07:16,536 –> 00:07:17,703 There are those 249 00:07:17,703 –> 00:07:19,539 not using AI yet today. 250 00:07:20,406 –> 00:07:22,141 I’m going to ask you, I’m guessing there’s probably a low 251 00:07:22,141 –> 00:07:23,342 number here, but 252 00:07:23,543 –> 00:07:25,011 how many of you in your organizations, 253 00:07:25,011 –> 00:07:26,279 they’re like, yeah, no, we’re not 254 00:07:26,279 –> 00:07:27,480 using AI. 255 00:07:27,547 –> 00:07:28,981 So hands anyone? 256 00:07:30,550 –> 00:07:31,551 Good. 257 00:07:31,551 –> 00:07:32,752 That’s great. 258 00:07:32,885 –> 00:07:34,854 There’s going to be certain things we talk about here 259 00:07:34,854 –> 00:07:36,456 for those that do operate that way. 260 00:07:36,756 –> 00:07:38,257 You want to make sure these things are in place. 261 00:07:38,257 –> 00:07:39,725 Because we’re going to show kind of a progression 262 00:07:39,725 –> 00:07:41,427 for governance as you work through this. 263 00:07:41,828 –> 00:07:42,695 So hopefully those 264 00:07:42,695 –> 00:07:43,896 those things are helpful. 265 00:07:44,297 –> 00:07:45,031 How many of you 266 00:07:45,031 –> 00:07:46,732 your organization allows you to use 267 00:07:47,066 –> 00:07:48,334 free public AI, 268 00:07:48,334 –> 00:07:49,669 but they’re not investing in things 269 00:07:49,669 –> 00:07:50,903 like CoPilot and not building 270 00:07:50,903 –> 00:07:52,104 agents yet. 271 00:07:52,405 –> 00:07:53,606 So hands. 272 00:07:54,040 –> 00:07:54,440 Okay. 273 00:07:54,440 –> 00:07:55,641 Thank you. 274 00:07:56,242 –> 00:07:57,543 How many of your organizations 275 00:07:57,543 –> 00:08:00,646 are investing in AI systems 276 00:08:00,646 –> 00:08:02,882 like CoPilot licensed for CoPilot? 277 00:08:03,749 –> 00:08:04,951 Great. Great. 278 00:08:05,218 –> 00:08:05,885 Changed a lot. 279 00:08:05,885 –> 00:08:07,787 I left Microsoft two years ago. 280 00:08:08,354 –> 00:08:08,588 Three. 281 00:08:08,588 –> 00:08:10,490 Now, I guess in 2023, 282 00:08:10,490 –> 00:08:11,757 CoPilot is just coming out. 283 00:08:12,158 –> 00:08:13,359 I can tell you internally, 284 00:08:13,759 –> 00:08:14,961 everyone there is like, 285 00:08:15,127 –> 00:08:16,462 what is this thing? Right? 286 00:08:16,496 –> 00:08:17,897 We’re they’re struggling 287 00:08:17,897 –> 00:08:19,799 to use it ourselves, right? 288 00:08:20,266 –> 00:08:21,334 Come a long way. 289 00:08:21,334 –> 00:08:23,402 In three years, I have not seen a technology 290 00:08:23,402 –> 00:08:25,271 get adapt adopted this quickly. 291 00:08:25,671 –> 00:08:27,306 In my 30 years in the business. 292 00:08:29,375 –> 00:08:29,709 The next 293 00:08:29,709 –> 00:08:30,910 part of this is how many of you 294 00:08:30,910 –> 00:08:32,378 are actually building your own 295 00:08:32,378 –> 00:08:33,579 AI offerings. 296 00:08:33,779 –> 00:08:34,947 This could be more CoPilot 297 00:08:34,947 –> 00:08:37,416 Studio, could be AI Foundry could be host. 298 00:08:37,416 –> 00:08:38,985 All right. Thank you. Thank you. 299 00:08:39,385 –> 00:08:40,686 When you take that on, it’s a whole 300 00:08:40,686 –> 00:08:42,989 nother level of responsibility when you’re doing this, 301 00:08:43,222 –> 00:08:44,490 because you’re now providing a product 302 00:08:44,490 –> 00:08:45,691 for other people to use. 303 00:08:45,958 –> 00:08:46,826 So you have to take on 304 00:08:46,826 –> 00:08:49,428 some of the things that Microsoft has to take on itself 305 00:08:49,428 –> 00:08:50,630 or other vendors 306 00:08:50,763 –> 00:08:52,398 when they’re delivering an AI product. 307 00:08:52,565 –> 00:08:54,166 So we’re going to talk about all that as we go. 308 00:08:55,801 –> 00:08:57,003 First of all, 309 00:08:57,103 –> 00:08:58,371 if you’re not using AI directly, 310 00:08:58,371 –> 00:09:00,006 and again, I appreciate most of you aren’t. 311 00:09:00,806 –> 00:09:01,507 Thing to remember 312 00:09:01,507 –> 00:09:03,609 there is that there’s no avoiding AI. 313 00:09:04,110 –> 00:09:05,511 It’s here to stay. There is. 314 00:09:05,912 –> 00:09:07,146 You can’t put your head in the sand 315 00:09:07,146 –> 00:09:08,514 and hope it bypasses you. 316 00:09:08,748 –> 00:09:09,815 You have to embrace it. 317 00:09:09,815 –> 00:09:10,983 You have to realize that this is 318 00:09:10,983 –> 00:09:12,919 this is our new reality, right? 319 00:09:13,686 –> 00:09:14,387 In this stage, 320 00:09:14,387 –> 00:09:15,721 we encourage a lot of organizations 321 00:09:15,721 –> 00:09:16,923 creating the AI policy. 322 00:09:17,390 –> 00:09:18,824 If you don’t have one, it’s a good time 323 00:09:18,824 –> 00:09:20,226 to kind of get caught up on that. 324 00:09:20,593 –> 00:09:22,028 So you can create that policy for 325 00:09:22,028 –> 00:09:23,529 how do you want to responsibly use 326 00:09:23,529 –> 00:09:24,897 AI within your organization? 327 00:09:25,464 –> 00:09:26,666 Right here 328 00:09:26,766 –> 00:09:27,700 we highlighted here 329 00:09:27,700 –> 00:09:30,202 because for organizations that haven’t quite gotten there 330 00:09:30,202 –> 00:09:32,705 yet, you’re trying to create that vision for okay, 331 00:09:32,705 –> 00:09:33,773 how would we use this. 332 00:09:33,773 –> 00:09:34,807 You got to get there. 333 00:09:34,807 –> 00:09:35,975 So how do you create that plan 334 00:09:35,975 –> 00:09:37,810 for what you would allow use for? 335 00:09:38,978 –> 00:09:39,612 Other important 336 00:09:39,612 –> 00:09:41,547 part here is just educating yourself. 337 00:09:42,114 –> 00:09:43,215 A lot of you probably have. 338 00:09:43,215 –> 00:09:45,217 Being here today is probably helped a little bit, 339 00:09:45,585 –> 00:09:47,086 but you’ve probably gone through a lot of what 340 00:09:47,286 –> 00:09:48,621 we’re going to talk about there already. 341 00:09:50,022 –> 00:09:51,524 The other thing there is to restrict 342 00:09:51,524 –> 00:09:53,159 AI use an access until you’re ready. 343 00:09:53,626 –> 00:09:53,859 Right. 344 00:09:53,859 –> 00:09:55,328 If you’re really trying to avoid that, 345 00:09:55,728 –> 00:09:57,630 there are some tools there which can be used 346 00:09:57,930 –> 00:09:58,898 further as you’re starting 347 00:09:58,898 –> 00:10:00,099 to control the environment 348 00:10:00,466 –> 00:10:01,467 that are beneficial here 349 00:10:01,467 –> 00:10:02,668 to to put in place. 350 00:10:03,669 –> 00:10:06,739 Now, I appreciate that after lunch. 351 00:10:07,406 –> 00:10:08,874 Probably not the most exciting 352 00:10:08,874 –> 00:10:10,910 topic in the world, right? 353 00:10:11,277 –> 00:10:13,813 So we put in some fun videos here. 354 00:10:13,813 –> 00:10:15,414 We had a couple of our associates 355 00:10:15,748 –> 00:10:17,316 play our some fictional characters, 356 00:10:17,316 –> 00:10:18,517 Daryl and Daryl, 357 00:10:18,517 –> 00:10:19,986 and highlighting some, 358 00:10:19,986 –> 00:10:22,355 some caricature style, 359 00:10:22,688 –> 00:10:25,257 events that people 360 00:10:25,658 –> 00:10:28,127 might, might be experience as they’re going through this. 361 00:10:29,695 –> 00:10:30,997 So here’s Daryl and Daryl 362 00:10:31,430 –> 00:10:32,398 struggling with 363 00:10:32,398 –> 00:10:34,934 they’re not necessarily using AI yet, 364 00:10:35,501 –> 00:10:36,369 but it’s it highlights 365 00:10:36,369 –> 00:10:37,737 the importance of AI awareness. 366 00:10:39,905 –> 00:10:40,673 Hey, Daryl. 367 00:10:40,673 –> 00:10:42,842 Did you catch the game last night? 368 00:10:42,842 –> 00:10:43,609 Kansas game? 369 00:10:43,609 –> 00:10:44,610 No, I didn’t. No. 370 00:10:44,610 –> 00:10:45,811 Come on, come on, come on. 371 00:10:45,978 –> 00:10:47,146 Look at this. 372 00:10:47,146 –> 00:10:48,314 Have you ever seen 373 00:10:48,314 –> 00:10:49,548 a baseball game like this before? 374 00:10:49,582 –> 00:10:50,182 Look at this video. 375 00:10:50,182 –> 00:10:51,384 Check this out. 376 00:10:52,618 –> 00:10:53,953 What is going on? 377 00:10:53,953 –> 00:10:55,888 I just want to say I can’t stop watching this. 378 00:10:55,888 –> 00:10:57,390 Like, what is that video? 379 00:10:57,523 –> 00:10:58,824 You sure this is real? 380 00:10:58,891 –> 00:11:00,092 It’s a video, Darryl. 381 00:11:00,159 –> 00:11:01,360 Videos are real. 382 00:11:03,029 –> 00:11:03,696 Darryl. 383 00:11:03,696 –> 00:11:06,065 This is clearly AI generated. 384 00:11:06,232 –> 00:11:07,500 But it’s a video. 385 00:11:07,533 –> 00:11:10,503 No no no no no, that does not happen 386 00:11:10,503 –> 00:11:13,239 in basketball games, I clearly cannot. 387 00:11:13,239 –> 00:11:15,474 You’re saying I can do this? 388 00:11:15,508 –> 00:11:17,510 Yes, I can do a lot of things, Darryl. 389 00:11:17,510 –> 00:11:18,477 And this is one of them. 390 00:11:20,046 –> 00:11:20,646 You got to watch out 391 00:11:20,646 –> 00:11:21,847 for these things. 392 00:11:23,416 –> 00:11:23,649 Thing 393 00:11:23,649 –> 00:11:25,918 you’re highlighting there is that even if you’re not using AI, 394 00:11:25,951 –> 00:11:27,687 you need to ensure that you understand 395 00:11:27,687 –> 00:11:28,888 what I can do 396 00:11:29,221 –> 00:11:31,057 and be able to police yourself 397 00:11:31,057 –> 00:11:33,192 in terms of trusting what you’re seeing out there. 398 00:11:33,359 –> 00:11:34,360 Right then. 399 00:11:34,360 –> 00:11:36,362 This goes for whatever stage you’re in. 400 00:11:36,362 –> 00:11:36,662 In this. 401 00:11:36,662 –> 00:11:38,431 It’s an important aspect of this. 402 00:11:38,664 –> 00:11:40,266 You need to understand that there’s 403 00:11:40,266 –> 00:11:42,601 a lot of possibilities with AI. 404 00:11:42,935 –> 00:11:44,804 Deepfakes is a very real thing. 405 00:11:44,804 –> 00:11:47,106 The phishing, sophistication 406 00:11:47,106 –> 00:11:49,809 has gone through the roof now with AI there. 407 00:11:49,809 –> 00:11:51,677 Those emails you’ll get from, 408 00:11:51,677 –> 00:11:53,646 you know, that uncle in a foreign country 409 00:11:54,280 –> 00:11:57,083 look far more legitimate than they used to, right? 410 00:11:57,083 –> 00:11:57,483 Used to. 411 00:11:57,483 –> 00:11:59,418 We’ve had been dealing with these things for a long time, 412 00:11:59,719 –> 00:12:00,953 and just making that more, 413 00:12:01,587 –> 00:12:04,356 makes it harder to to discern. 414 00:12:04,356 –> 00:12:04,924 Right. 415 00:12:04,924 –> 00:12:05,958 But there are things you can do 416 00:12:05,958 –> 00:12:07,426 to educate yourself 417 00:12:07,426 –> 00:12:08,627 to, to deal with that. 418 00:12:08,794 –> 00:12:10,029 It’s important that everybody does 419 00:12:11,897 –> 00:12:12,665 in terms of, 420 00:12:12,665 –> 00:12:15,234 kind of getting ready for AI 421 00:12:15,468 –> 00:12:17,203 and or if you’re already in AI, 422 00:12:17,570 –> 00:12:19,338 I highly encourage you 423 00:12:19,772 –> 00:12:20,973 to investigate 424 00:12:21,240 –> 00:12:23,209 embracing a zero trust strategy. 425 00:12:23,843 –> 00:12:24,677 Zero trust. 426 00:12:24,677 –> 00:12:27,113 And the reason I really, really emphasize 427 00:12:27,113 –> 00:12:29,115 this is zero trust. 428 00:12:29,115 –> 00:12:30,616 Even though it’s a security model, 429 00:12:30,850 –> 00:12:32,885 security is usually about saying no 430 00:12:33,786 –> 00:12:34,086 zero. 431 00:12:34,086 –> 00:12:35,855 Trust is not zero. 432 00:12:35,855 –> 00:12:37,957 Trust is actually about saying yes. 433 00:12:38,257 –> 00:12:39,458 How do you enable 434 00:12:39,925 –> 00:12:41,927 your organization and the security 435 00:12:42,061 –> 00:12:43,429 within your organization to be able to say 436 00:12:43,429 –> 00:12:44,630 yes to things? 437 00:12:44,797 –> 00:12:45,998 This is a way to do that. 438 00:12:46,031 –> 00:12:47,666 You have a strong enough security model. 439 00:12:47,933 –> 00:12:49,835 It allows security become an enabler 440 00:12:50,236 –> 00:12:51,971 of certain things you want to do. 441 00:12:52,571 –> 00:12:53,773 If you’re getting started with this, 442 00:12:53,839 –> 00:12:54,940 I highly recommend 443 00:12:54,940 –> 00:12:56,275 running a zero trust assessment 444 00:12:56,275 –> 00:12:57,476 on your environment. 445 00:12:57,510 –> 00:12:58,511 Maybe running a workshop 446 00:12:58,511 –> 00:12:59,979 to get familiar with the concept. 447 00:13:00,446 –> 00:13:02,214 It’s important to understand 448 00:13:02,214 –> 00:13:04,016 the real philosophy of that. 449 00:13:04,016 –> 00:13:05,584 And it’s not about, you know, 450 00:13:05,584 –> 00:13:07,419 never trust anybody in that, which it is. 451 00:13:07,419 –> 00:13:09,355 But there’s a reason for that. 452 00:13:09,355 –> 00:13:11,690 And the purpose of that is that if I treat it that way, 453 00:13:12,057 –> 00:13:13,926 I treat it as kind of assume breach. 454 00:13:14,593 –> 00:13:15,528 That allows me to, 455 00:13:16,595 –> 00:13:18,197 to lock things down enough 456 00:13:18,798 –> 00:13:20,666 that I’m able to enable 457 00:13:20,666 –> 00:13:21,867 other ways of working. 458 00:13:21,867 –> 00:13:23,402 This was highly important as well. 459 00:13:23,402 –> 00:13:25,437 In 2020, we went to hybrid work models, 460 00:13:25,838 –> 00:13:26,939 a lot of organizations. 461 00:13:26,939 –> 00:13:28,207 This philosophy really 462 00:13:28,207 –> 00:13:29,975 help them out because they’re like, 463 00:13:29,975 –> 00:13:31,644 I’ve got to be able to enable these people. 464 00:13:31,844 –> 00:13:33,612 How do I secure the core enough 465 00:13:34,079 –> 00:13:36,081 that I can allow people to operate 466 00:13:36,081 –> 00:13:37,550 outside of the network perimeters? 467 00:13:37,550 –> 00:13:41,520 Perhaps I operate in, identities 468 00:13:41,821 –> 00:13:43,155 very important for this. 469 00:13:44,190 –> 00:13:45,825 So I would encourage you all to make progress 470 00:13:45,825 –> 00:13:47,026 towards adoption of that. 471 00:13:47,393 –> 00:13:47,760 Your thing. 472 00:13:47,760 –> 00:13:48,961 Do make sure you have an acceptable 473 00:13:48,994 –> 00:13:50,196 AI usage policy. 474 00:13:50,629 –> 00:13:51,363 Make sure you’ve set 475 00:13:51,363 –> 00:13:52,665 some kind of parameters 476 00:13:52,932 –> 00:13:53,999 for your organization in terms 477 00:13:53,999 –> 00:13:55,868 of what we allow for what we don’t 478 00:13:56,502 –> 00:13:57,803 in terms of AI use. 479 00:13:58,571 –> 00:13:58,904 Right. 480 00:13:59,939 –> 00:14:02,107 I would highly encourage you to use 481 00:14:02,107 –> 00:14:03,475 don’t just have this be it 482 00:14:03,576 –> 00:14:05,411 and security bring H.R. 483 00:14:05,411 –> 00:14:06,712 into this, bring legal under this, 484 00:14:07,012 –> 00:14:08,914 bring as many people from your organization as you want. 485 00:14:08,914 –> 00:14:10,683 But I certainly would start with those four. 486 00:14:11,083 –> 00:14:12,284 You don’t want to muddy the waters 487 00:14:12,284 –> 00:14:13,686 so much you can’t get anything done, 488 00:14:13,986 –> 00:14:14,887 but you need certain core 489 00:14:14,887 –> 00:14:16,088 representation 490 00:14:16,188 –> 00:14:17,389 in order to do this right. 491 00:14:20,626 –> 00:14:21,427 Education level 492 00:14:21,427 –> 00:14:23,462 and understanding what I can do. 493 00:14:24,029 –> 00:14:25,931 Identifying AI generated content. 494 00:14:26,232 –> 00:14:27,800 There’s a host of online 495 00:14:28,100 –> 00:14:29,001 learnings on this. 496 00:14:29,001 –> 00:14:31,604 I would encourage anybody that hasn’t been familiar. 497 00:14:31,637 –> 00:14:33,272 This this is an ongoing thing. 498 00:14:33,539 –> 00:14:34,940 Things are changing all the time. 499 00:14:34,940 –> 00:14:36,609 Make sure the organizations educated 500 00:14:37,576 –> 00:14:38,310 first barrier 501 00:14:38,310 –> 00:14:39,578 there is making sure people are doing 502 00:14:39,578 –> 00:14:42,014 smart things and not making stupid mistakes right? 503 00:14:45,150 –> 00:14:46,719 In terms of restricting AI 504 00:14:46,719 –> 00:14:49,188 tool usage, we highlight this here. 505 00:14:49,688 –> 00:14:50,856 You can use the things 506 00:14:50,856 –> 00:14:52,157 like Defender for Cloud Apps. 507 00:14:52,591 –> 00:14:53,525 This actually feeds 508 00:14:53,525 –> 00:14:54,727 off of your firewalls 509 00:14:54,727 –> 00:14:56,161 and the information within there. 510 00:14:56,662 –> 00:14:57,630 It’s a great tool 511 00:14:57,630 –> 00:15:00,132 for being able to start examining 512 00:15:00,366 –> 00:15:01,934 what people are doing with AI 513 00:15:02,434 –> 00:15:03,569 and offering 514 00:15:03,569 –> 00:15:05,004 a viewpoint on that 515 00:15:05,004 –> 00:15:06,205 in terms of whether that’s 516 00:15:06,205 –> 00:15:07,940 sanctioned or unsanctioned. 517 00:15:08,474 –> 00:15:09,775 You market is unsanctioned. 518 00:15:10,175 –> 00:15:12,077 It’ll start putting alerts out there. 519 00:15:12,578 –> 00:15:14,046 If you have defender for endpoints in there, 520 00:15:14,179 –> 00:15:15,381 you can actually block use. 521 00:15:16,148 –> 00:15:17,583 You can also do this through certain 522 00:15:17,583 –> 00:15:18,784 networking tools as well. 523 00:15:19,084 –> 00:15:20,352 But as important as that 524 00:15:20,352 –> 00:15:23,789 list of AI agents and tools out there 525 00:15:23,789 –> 00:15:24,990 continues to grow, 526 00:15:25,057 –> 00:15:26,325 not all of which can be trusted. 527 00:15:26,926 –> 00:15:28,127 The nice thing is that Microsoft’s 528 00:15:28,127 –> 00:15:29,361 keeping an eye on that for you. 529 00:15:29,595 –> 00:15:30,829 They’re providing risk scoring 530 00:15:30,896 –> 00:15:32,197 in terms of the tooling 531 00:15:32,197 –> 00:15:33,399 and agent that are out there. 532 00:15:33,499 –> 00:15:34,900 Just provides a guide for everybody. 533 00:15:34,934 –> 00:15:36,368 See, don’t have to be master of that domain 534 00:15:36,368 –> 00:15:39,071 to top of everything else you’re doing in your organization. 535 00:15:40,839 –> 00:15:42,041 All right. 536 00:15:42,441 –> 00:15:43,642 Second stage 537 00:15:44,076 –> 00:15:45,277 using public AI. 538 00:15:45,811 –> 00:15:46,078 Right. 539 00:15:46,078 –> 00:15:48,480 So here I’m using things like ChatGPT 540 00:15:49,214 –> 00:15:52,451 could be, CoPilot chat could be, 541 00:15:52,451 –> 00:15:54,520 Claude, a host of other, 542 00:15:54,553 –> 00:15:55,754 other tools. 543 00:15:56,388 –> 00:15:57,990 First thing to be aware of here is be aware 544 00:15:57,990 –> 00:15:59,458 of what you share. Right. 545 00:15:59,458 –> 00:16:01,026 And what’s happening with that information. 546 00:16:01,493 –> 00:16:02,361 As you’re sharing, 547 00:16:02,361 –> 00:16:03,662 even as you’re prompting, 548 00:16:03,929 –> 00:16:05,464 you are sharing information 549 00:16:05,464 –> 00:16:06,665 just in those prompts. 550 00:16:06,899 –> 00:16:08,000 If you share data up 551 00:16:08,000 –> 00:16:09,201 there, share documents, 552 00:16:09,668 –> 00:16:10,836 you’re investing a whole 553 00:16:10,836 –> 00:16:13,038 nother level of information out there 554 00:16:13,238 –> 00:16:15,341 to be used by whatever tool. 555 00:16:15,341 –> 00:16:17,176 And whoever’s running that to on the back end. 556 00:16:18,010 –> 00:16:19,411 Continue to refine your policy 557 00:16:19,411 –> 00:16:20,612 and educate here. 558 00:16:21,246 –> 00:16:22,748 Make sure you’re protecting 559 00:16:22,748 –> 00:16:25,217 your organization against the untrusted tooling. 560 00:16:25,517 –> 00:16:27,119 This is evolving things like Defender 561 00:16:27,119 –> 00:16:28,320 for Cloud Apps 562 00:16:28,320 –> 00:16:29,955 to allow for trusted tooling, 563 00:16:29,955 –> 00:16:31,190 but maybe not the untrusted 564 00:16:32,725 –> 00:16:33,959 let’s have another video in terms of 565 00:16:33,959 –> 00:16:35,394 an example of where this can go wrong. 566 00:16:36,695 –> 00:16:38,063 Daryl, come over here, I want to show you something. 567 00:16:38,130 –> 00:16:39,298 Yea, what you working on Daryl? 568 00:16:39,298 –> 00:16:42,034 I am just doing some analysis 569 00:16:42,034 –> 00:16:43,235 for the sales team. 570 00:16:43,469 –> 00:16:44,403 And, 571 00:16:44,403 –> 00:16:46,372 there’s this new Russian AI tool. 572 00:16:46,405 –> 00:16:46,839 Yeah. 573 00:16:46,839 –> 00:16:48,674 It’s supposed to give you tremendous outputs 574 00:16:48,674 –> 00:16:49,942 just for the switch 575 00:16:50,309 –> 00:16:51,510 data prior to security. 576 00:16:51,643 –> 00:16:52,611 Nothing to worry about. 577 00:16:52,611 –> 00:16:53,812 It’s really cool. 578 00:16:53,879 –> 00:16:55,180 So are we paying for yours? 579 00:16:55,347 –> 00:16:56,382 Oh, no. Completely free. 580 00:16:56,382 –> 00:16:57,316 Oh, sweet. 581 00:16:57,316 –> 00:16:58,550 It’s the goodness of their heart. 582 00:16:58,984 –> 00:17:01,653 So can I, let’s go ahead here. 583 00:17:01,653 –> 00:17:04,056 I’m going to upload the sales 584 00:17:04,056 –> 00:17:05,624 forecast in spreadsheets. 585 00:17:05,691 –> 00:17:07,493 There’s not really much in it other than just, like, 586 00:17:07,760 –> 00:17:08,827 our client’s names 587 00:17:08,827 –> 00:17:10,095 and their addresses. Now. 588 00:17:10,129 –> 00:17:11,163 Yeah, it’s stuff like that. 589 00:17:11,163 –> 00:17:12,598 Maybe I should send you mine, too. 590 00:17:12,598 –> 00:17:13,999 And then you could do them both at once. 591 00:17:14,266 –> 00:17:15,401 That would be great. 592 00:17:15,401 –> 00:17:16,335 He should. Okay. 593 00:17:16,335 –> 00:17:19,004 I’m gonna build my clients out. I’ve got all my customers here. 594 00:17:19,171 –> 00:17:20,139 Here we go. Yeah. 595 00:17:20,139 –> 00:17:21,006 Throw that in there. 596 00:17:21,006 –> 00:17:23,475 OK, cool to start it. 597 00:17:23,675 –> 00:17:26,345 Catch that one too. Submit. 598 00:17:28,247 –> 00:17:28,614 All right. 599 00:17:28,614 –> 00:17:29,882 Why is my computer black? 600 00:17:30,182 –> 00:17:31,383 Wait. Daryl? 601 00:17:31,650 –> 00:17:32,584 Yeah. 602 00:17:32,584 –> 00:17:34,053 Yeah, my computer’s black. 603 00:17:35,654 –> 00:17:36,855 Me too. 604 00:17:37,456 –> 00:17:38,657 Mine too. 605 00:17:38,857 –> 00:17:39,825 Oh, no. 606 00:17:39,825 –> 00:17:41,026 We’ve been hacked. 607 00:17:41,160 –> 00:17:41,927 What happened? 608 00:17:41,927 –> 00:17:43,128 What site was that? 609 00:17:43,262 –> 00:17:44,463 Oh, God. 610 00:17:47,766 –> 00:17:48,967 Oh, Daryl. 611 00:17:50,335 –> 00:17:50,803 so. Yeah. 612 00:17:50,803 –> 00:17:52,938 So again, extreme example, 613 00:17:52,938 –> 00:17:54,373 but you get the point right there. 614 00:17:54,406 –> 00:17:55,774 There are risks to sharing 615 00:17:55,774 –> 00:17:57,142 information through public AI. 616 00:17:57,376 –> 00:17:58,343 You want to make sure you’re using 617 00:17:58,343 –> 00:17:59,878 trusted tooling tooling 618 00:17:59,878 –> 00:18:01,313 that’s not sharing your data 619 00:18:01,313 –> 00:18:03,615 outside of the sessions that you’re running in. 620 00:18:03,715 –> 00:18:04,850 Right. 621 00:18:04,850 –> 00:18:05,651 CoPilot chat 622 00:18:05,651 –> 00:18:07,286 does a nice job of this, right? 623 00:18:07,486 –> 00:18:09,388 Not all tools out there are created the same. 624 00:18:09,388 –> 00:18:11,490 Some will mine your data that you’re sharing with them 625 00:18:11,757 –> 00:18:13,559 may not lead to hacking incidents, 626 00:18:13,559 –> 00:18:14,760 but it will lead to, 627 00:18:15,094 –> 00:18:16,028 data exfiltration 628 00:18:16,028 –> 00:18:17,729 that you may not have anticipated. 629 00:18:18,163 –> 00:18:20,833 So it’s important to continue to use 630 00:18:21,066 –> 00:18:22,634 trusted publicly AI tools. 631 00:18:22,634 –> 00:18:24,636 There’s a lot of readouts on that. 632 00:18:24,636 –> 00:18:26,572 There’s a host of of lists. 633 00:18:26,605 –> 00:18:28,173 And then people that can guide you into what 634 00:18:28,440 –> 00:18:30,042 what’s safe, what’s not and why. 635 00:18:31,677 –> 00:18:32,411 One important 636 00:18:32,411 –> 00:18:34,480 aspect here I want to highlight is that as you’re 637 00:18:34,480 –> 00:18:35,681 using AI, 638 00:18:36,515 –> 00:18:37,783 you want to treat that like 639 00:18:37,816 –> 00:18:40,219 you’re talking to a person, right? 640 00:18:40,219 –> 00:18:42,187 So if I’m talking to public AI, 641 00:18:42,621 –> 00:18:44,289 I want to treat it as if I’m talking to 642 00:18:44,289 –> 00:18:46,058 maybe somebody outside of my organization. 643 00:18:46,725 –> 00:18:47,960 If I’m asking questions 644 00:18:47,960 –> 00:18:49,161 or sharing information there, 645 00:18:49,328 –> 00:18:51,430 would I tell them that? Right. 646 00:18:51,430 –> 00:18:52,664 And being mindful of 647 00:18:52,664 –> 00:18:54,166 of my responsibilities to whoever 648 00:18:54,166 –> 00:18:55,367 I’m working for, 649 00:18:55,434 –> 00:18:57,069 it’s just kind of an initial good 650 00:18:57,069 –> 00:18:58,670 frame of mind to to have. 651 00:18:58,670 –> 00:19:00,105 They’re like, what I share 652 00:19:00,105 –> 00:19:01,373 this with with a competitor. 653 00:19:02,207 –> 00:19:03,242 Probably not. 654 00:19:03,242 –> 00:19:03,742 Maybe not. 655 00:19:03,742 –> 00:19:04,977 Want to put it in the public eye 656 00:19:05,477 –> 00:19:06,678 right. 657 00:19:06,945 –> 00:19:07,946 The other thing is that, 658 00:19:07,946 –> 00:19:10,749 you want to make sure to update your 659 00:19:10,749 –> 00:19:11,950 AI policy 660 00:19:12,050 –> 00:19:13,252 to kind of make sure you’re 661 00:19:13,418 –> 00:19:14,620 taking into account 662 00:19:14,887 –> 00:19:15,921 what you want to allow for 663 00:19:15,921 –> 00:19:18,023 as you’ve now entered this, this age, 664 00:19:18,023 –> 00:19:20,225 and define what those acceptable tools are. 665 00:19:20,526 –> 00:19:21,693 Make sure you’ve communicated 666 00:19:21,693 –> 00:19:22,895 that with everybody. 667 00:19:22,928 –> 00:19:24,096 It’s one thing just to say, okay, 668 00:19:24,096 –> 00:19:25,297 we’re just going to use this. 669 00:19:25,564 –> 00:19:26,298 You still have people 670 00:19:26,298 –> 00:19:27,766 going around, right? 671 00:19:28,000 –> 00:19:29,134 That’s not helpful. 672 00:19:29,134 –> 00:19:30,569 Need to make sure people are aware 673 00:19:30,569 –> 00:19:31,770 that this is available. 674 00:19:31,770 –> 00:19:34,139 This will help against the shadow IT piece. 675 00:19:34,406 –> 00:19:37,109 Or if you’re saying like, hey, we have these trusted tools to use, 676 00:19:37,442 –> 00:19:39,011 stop using everything else, 677 00:19:39,144 –> 00:19:40,546 whether you’re blocking them or not. 678 00:19:42,915 –> 00:19:45,150 Education enablement again enabled. 679 00:19:45,617 –> 00:19:46,618 Make sure you’re running everybody 680 00:19:46,618 –> 00:19:49,388 through an appropriate I use training model. 681 00:19:49,421 –> 00:19:49,588 Again. 682 00:19:49,588 –> 00:19:51,190 There’s a whole host of these out there. 683 00:19:51,557 –> 00:19:52,691 We can help you with that. 684 00:19:52,691 –> 00:19:54,526 But there’s there’s plenty online that you can 685 00:19:54,526 –> 00:19:56,562 you can use to to get through this. 686 00:19:57,196 –> 00:19:59,331 What trusted AI tools are appropriate 687 00:19:59,331 –> 00:20:00,532 use cases for it? 688 00:20:00,999 –> 00:20:02,201 Risks of data sharing. 689 00:20:02,201 –> 00:20:04,636 It’s important people start to understand that when you’re 690 00:20:04,636 –> 00:20:06,638 sharing things out there, what might happen with that data? 691 00:20:06,638 –> 00:20:08,574 Why is that a problem? Right? 692 00:20:08,574 –> 00:20:09,474 Not everybody gets that. 693 00:20:09,474 –> 00:20:11,543 A lot of people are trying to do the right thing all the time, 694 00:20:12,144 –> 00:20:14,479 but mistakes happen when they don’t know 695 00:20:14,479 –> 00:20:16,181 that they’re making errors 696 00:20:16,181 –> 00:20:17,382 in judgment, 697 00:20:18,016 –> 00:20:19,551 handle ethical issues. 698 00:20:20,152 –> 00:20:21,553 You know, it’s important 699 00:20:21,553 –> 00:20:23,155 that we’re we’re being responsible 700 00:20:23,155 –> 00:20:25,157 and how we behave out there, too, 701 00:20:25,157 –> 00:20:26,558 because sometimes those prompts, 702 00:20:26,959 –> 00:20:27,759 we’re asking things 703 00:20:27,759 –> 00:20:29,895 maybe we shouldn’t be in there 704 00:20:30,195 –> 00:20:31,096 that is that’s 705 00:20:31,096 –> 00:20:32,631 recorded, that’s available out there 706 00:20:32,631 –> 00:20:34,199 that can come back to bite you too. 707 00:20:34,633 –> 00:20:35,667 So make sure that 708 00:20:35,667 –> 00:20:36,902 that everyone’s behaving 709 00:20:36,902 –> 00:20:38,937 appropriately can treat it like a person. 710 00:20:39,338 –> 00:20:40,639 You wouldn’t say it to another person. 711 00:20:40,906 –> 00:20:42,407 It shouldn’t be done with AI either. 712 00:20:42,975 –> 00:20:44,676 Most of you, I’m not worried about this, but 713 00:20:44,943 –> 00:20:46,778 that has been a problem for some in the past. 714 00:20:47,813 –> 00:20:49,014 Make sure to. 715 00:20:49,681 –> 00:20:51,984 This is where we get to that trust 716 00:20:51,984 –> 00:20:53,185 but verify. 717 00:20:53,385 –> 00:20:55,687 I might still be at the never trust. 718 00:20:56,088 –> 00:20:57,656 Verify everything you see here. 719 00:20:58,357 –> 00:20:58,924 I prefer 720 00:20:58,924 –> 00:21:00,392 AI tools that give me the sources 721 00:21:00,392 –> 00:21:01,727 for where the information came from. 722 00:21:02,227 –> 00:21:04,162 It sped up my finding that it gave me 723 00:21:04,162 –> 00:21:05,530 a good readout of it, but 724 00:21:05,530 –> 00:21:06,732 I’m still checking the source. 725 00:21:06,898 –> 00:21:08,667 Make sure that that readout was correct. 726 00:21:08,967 –> 00:21:10,602 Make sure that’s a legitimate source, 727 00:21:11,203 –> 00:21:12,437 where that information came from. 728 00:21:14,406 –> 00:21:15,907 Protection monitoring again, 729 00:21:15,907 –> 00:21:17,476 evolving Defender for Cloud Apps 730 00:21:18,443 –> 00:21:18,677 here. 731 00:21:18,677 –> 00:21:20,178 We’re going to start getting into though 732 00:21:20,779 –> 00:21:22,681 if I’m going to be sharing data out 733 00:21:22,681 –> 00:21:24,316 and be able to upload a file, 734 00:21:24,316 –> 00:21:26,685 maybe to even CoPilot chat 735 00:21:26,685 –> 00:21:29,288 or Claude or ChatGPT, 736 00:21:30,656 –> 00:21:32,691 I want to start protecting my data 737 00:21:32,958 –> 00:21:34,226 at the document level, 738 00:21:35,193 –> 00:21:36,295 because there’s certain data there 739 00:21:36,295 –> 00:21:37,796 that I don’t want getting out 740 00:21:38,096 –> 00:21:39,331 right now. 741 00:21:39,865 –> 00:21:42,067 This is a process unto itself. 742 00:21:42,167 –> 00:21:44,503 We’re going to talk a little bit about data governance protection 743 00:21:44,503 –> 00:21:46,905 throughout this, because it is an important topic 744 00:21:47,472 –> 00:21:48,240 at this point. 745 00:21:48,240 –> 00:21:50,909 All we’re stating is that for the high, 746 00:21:50,942 –> 00:21:52,511 really highly sensitive things 747 00:21:52,844 –> 00:21:54,713 you think about marking those documents 748 00:21:54,946 –> 00:21:56,181 as sensitive, possibly 749 00:21:56,181 –> 00:21:57,482 putting some encryption on those 750 00:21:57,783 –> 00:21:59,184 so that only certain people 751 00:21:59,184 –> 00:22:00,385 are able to get to those. 752 00:22:00,552 –> 00:22:02,387 These are highly sensitive business documents. 753 00:22:02,387 –> 00:22:03,555 I want to start there. 754 00:22:03,555 –> 00:22:05,123 Please protect yourself against the things 755 00:22:05,123 –> 00:22:06,658 that are the biggest threat to your organization 756 00:22:07,526 –> 00:22:08,760 involved, that over time, 757 00:22:09,928 –> 00:22:11,129 the other thing to do is, 758 00:22:11,330 –> 00:22:13,432 manage your browsers and devices. 759 00:22:14,566 –> 00:22:15,834 There are a host of tools 760 00:22:15,834 –> 00:22:18,303 that Purview has in concert with Intune, 761 00:22:18,737 –> 00:22:19,638 probably others as 762 00:22:19,638 –> 00:22:20,906 well, where Microsoft shop. 763 00:22:20,906 –> 00:22:23,442 So I’m going to be talking a lot about the Microsoft tooling today. 764 00:22:23,442 –> 00:22:24,743 But the principle is the same. 765 00:22:25,644 –> 00:22:26,978 You want to embed 766 00:22:26,978 –> 00:22:29,114 some level of DLP into the browser 767 00:22:29,414 –> 00:22:31,083 so it can monitor what you’re doing. 768 00:22:31,083 –> 00:22:32,551 Most of these 769 00:22:32,551 –> 00:22:34,553 AI tools exist within a browser. 770 00:22:34,886 –> 00:22:35,921 I then can control 771 00:22:35,921 –> 00:22:37,155 if people are accessing ones 772 00:22:37,155 –> 00:22:39,558 I want or ones that I don’t write. 773 00:22:39,825 –> 00:22:41,693 I also will ultimately allow me to see 774 00:22:41,927 –> 00:22:43,428 what are they doing in there? 775 00:22:44,229 –> 00:22:45,464 Even in the trusted ones, 776 00:22:46,998 –> 00:22:48,667 if I am or the device in the Purview, 777 00:22:48,834 –> 00:22:50,035 I can see some other 778 00:22:50,235 –> 00:22:51,603 data flows within there too. 779 00:22:51,603 –> 00:22:54,005 And what’s happening with them, with my sensitive information 780 00:22:54,506 –> 00:22:55,207 inside of there 781 00:22:55,207 –> 00:22:56,508 as well, just provides 782 00:22:56,508 –> 00:22:58,810 a lot of good visibility here, helps me evolve 783 00:22:58,810 –> 00:23:00,145 my policies, helps me 784 00:23:00,145 –> 00:23:01,546 have all my protection mechanisms. 785 00:23:01,847 –> 00:23:03,048 It’s a great place to start. 786 00:23:03,281 –> 00:23:05,283 Pretty low lift, to be honest, right? 787 00:23:05,283 –> 00:23:08,387 This is not a difficult thing to do, but it is important. 788 00:23:09,788 –> 00:23:10,522 The other thing I would do 789 00:23:10,522 –> 00:23:11,757 in this ties in with being able 790 00:23:11,757 –> 00:23:13,458 to monitor this from the DLP 791 00:23:14,025 –> 00:23:15,827 is be able to turn on Data 792 00:23:15,827 –> 00:23:17,896 Security Posture Management for for AI. 793 00:23:18,797 –> 00:23:19,431 This is something 794 00:23:19,431 –> 00:23:21,466 that’s part of the Purview E5 suite. 795 00:23:21,967 –> 00:23:24,035 It’s a great visibility 796 00:23:24,035 –> 00:23:26,071 tool, provides AI observability in there 797 00:23:26,505 –> 00:23:27,539 and you can start seeing 798 00:23:27,539 –> 00:23:28,740 what’s happening. 799 00:23:29,007 –> 00:23:31,209 Use this in concert with Insider Risk 800 00:23:31,510 –> 00:23:32,878 to create policies, 801 00:23:32,878 –> 00:23:34,846 which is another part of of Purview. 802 00:23:35,680 –> 00:23:37,349 And use this to create policies 803 00:23:37,349 –> 00:23:39,718 to monitor for certain generative AI use. 804 00:23:40,218 –> 00:23:41,052 And you can see kind of the 805 00:23:41,052 –> 00:23:42,354 what’s being shared on prompts, 806 00:23:42,788 –> 00:23:43,922 the information being shared in 807 00:23:43,922 –> 00:23:45,757 that nothing else is just providing 808 00:23:45,757 –> 00:23:46,958 observability at this point. 809 00:23:46,958 –> 00:23:49,127 So you can start to say, okay, we’ve got to tune this 810 00:23:49,127 –> 00:23:50,061 a little bit. 811 00:23:50,061 –> 00:23:51,596 I don’t really like this happening. 812 00:23:51,596 –> 00:23:52,764 Being able to manage that. 813 00:23:55,267 –> 00:23:55,967 That’s public 814 00:23:55,967 –> 00:23:56,735 AI. 815 00:23:56,735 –> 00:23:58,370 Let’s start getting into when we’re starting to pay for this 816 00:23:58,370 –> 00:23:59,571 a little more 817 00:24:00,405 –> 00:24:01,506 using AI systems. 818 00:24:01,506 –> 00:24:03,275 One, obviously use something 819 00:24:03,275 –> 00:24:05,410 that you trust the don’t pay for things you don’t trust. 820 00:24:05,410 –> 00:24:07,245 That’s just common sense. 821 00:24:08,313 –> 00:24:10,215 But so you’re using something 822 00:24:10,215 –> 00:24:11,416 like CoPilot. 823 00:24:11,416 –> 00:24:12,918 Be aware what you’re exposing there. 824 00:24:13,251 –> 00:24:14,786 Not so much to the outside world. 825 00:24:14,786 –> 00:24:16,588 CoPilot is going to protect you for a lot of that. 826 00:24:17,055 –> 00:24:17,889 This is what you’re sharing 827 00:24:17,889 –> 00:24:19,658 internally, right? 828 00:24:20,058 –> 00:24:21,159 There are a lot of threats 829 00:24:21,159 –> 00:24:22,461 that start to get exposed. 830 00:24:22,828 –> 00:24:24,229 If you don’t manage your environment 831 00:24:24,229 –> 00:24:26,264 properly and open up CoPilot to it, 832 00:24:26,731 –> 00:24:27,499 you’re going to expose 833 00:24:27,499 –> 00:24:29,301 all kinds of things that you never thought 834 00:24:29,301 –> 00:24:31,436 somebody would find or see 835 00:24:31,803 –> 00:24:32,771 or anything like that, 836 00:24:32,771 –> 00:24:35,474 because CoPilot searching all that and it’s it’s 837 00:24:35,540 –> 00:24:36,975 surfacing those in ways you wouldn’t 838 00:24:36,975 –> 00:24:38,176 even anticipate. 839 00:24:38,477 –> 00:24:39,077 So it’s important 840 00:24:39,077 –> 00:24:40,412 to put some protections on that 841 00:24:41,279 –> 00:24:42,948 against that unexpected 842 00:24:42,948 –> 00:24:45,250 information sharing internally. 843 00:24:46,384 –> 00:24:47,886 So you an example what can happen. 844 00:24:50,689 –> 00:24:51,156 Hey, Daryl. 845 00:24:51,156 –> 00:24:52,724 Did you hear the boss 846 00:24:52,724 –> 00:24:53,925 story about Project Phoenix? 847 00:24:54,192 –> 00:24:55,160 I haven’t heard of that. No. 848 00:24:55,160 –> 00:24:57,062 You want to put it in CoPilot and see if we get anything? 849 00:24:57,562 –> 00:24:57,896 I just. 850 00:24:57,896 –> 00:24:58,864 Oh, I’d love to know more. 851 00:24:58,864 –> 00:25:00,065 Yeah. 852 00:25:00,332 –> 00:25:01,766 Ooh. CoPilot’s got some hits. 853 00:25:02,567 –> 00:25:04,202 I was presenting information. 854 00:25:04,202 –> 00:25:06,771 Project Phoenix is a strategic acquisition. 855 00:25:07,539 –> 00:25:08,740 Acquisition? 856 00:25:08,740 –> 00:25:09,574 Wait wait wait. 857 00:25:09,574 –> 00:25:11,309 Are we being acquired 858 00:25:11,309 –> 00:25:13,578 or are we acquiring? CoPilot, 859 00:25:13,612 –> 00:25:16,081 what is the new company org chart? 860 00:25:16,448 –> 00:25:17,482 Wait, this acquisition 861 00:25:17,482 –> 00:25:18,950 is going to close next week. 862 00:25:19,351 –> 00:25:20,385 Nice. Alright. 863 00:25:20,385 –> 00:25:21,219 Sweet. 864 00:25:21,219 –> 00:25:22,420 Could be a big payday. 865 00:25:23,121 –> 00:25:24,322 Stock? 866 00:25:24,456 –> 00:25:26,224 Daryl, wait. 867 00:25:27,092 –> 00:25:28,293 We’re not in this. 868 00:25:28,393 –> 00:25:29,194 You’re not in this 869 00:25:29,194 –> 00:25:30,195 org chart. 870 00:25:30,195 –> 00:25:31,863 No. There must be a mistake. 871 00:25:31,863 –> 00:25:32,998 I’m not. I’m not. 872 00:25:32,998 –> 00:25:35,767 And this is straight from our boss. 873 00:25:35,800 –> 00:25:37,168 I’m not in this org chart either 874 00:25:38,069 –> 00:25:39,037 875 00:25:40,171 –> 00:25:41,673 Are we 876 00:25:42,841 –> 00:25:44,042 We’re not going to get fired, are we? 877 00:25:44,042 –> 00:25:45,243 But I. 878 00:25:46,444 –> 00:25:47,045 This is it. 879 00:25:47,045 –> 00:25:48,747 This is ridiculous. God! 880 00:25:51,016 –> 00:25:52,050 And now you have an HR problem 881 00:25:52,050 –> 00:25:53,618 that didn’t have to happen, right? 882 00:25:53,785 –> 00:25:54,653 Like that’s. 883 00:25:54,653 –> 00:25:55,787 This is why it’s important 884 00:25:55,787 –> 00:25:57,756 to be careful of what you’re sharing internally. 885 00:25:58,290 –> 00:25:59,891 Innocent enough. Right? 886 00:25:59,891 –> 00:26:01,359 They’re not doing anything wrong. 887 00:26:01,359 –> 00:26:02,561 They’re curious, 888 00:26:02,661 –> 00:26:04,763 but they don’t have any business knowing that given where they’re 889 00:26:04,763 –> 00:26:06,264 at in the organization right now, that’s 890 00:26:06,264 –> 00:26:07,899 there are reasons that that’s managed 891 00:26:08,033 –> 00:26:09,200 right. 892 00:26:09,334 –> 00:26:09,968 This is why 893 00:26:09,968 –> 00:26:12,304 data governance and protection is so important. 894 00:26:12,404 –> 00:26:15,006 When you enter the world, particularly of agenetic 895 00:26:15,006 –> 00:26:16,207 AI systems 896 00:26:16,274 –> 00:26:17,609 it’s going to expose data 897 00:26:17,642 –> 00:26:20,845 out there that you had no idea was there still. 898 00:26:21,112 –> 00:26:21,713 Right. 899 00:26:21,713 –> 00:26:23,481 Or it’s going to expose 900 00:26:23,481 –> 00:26:24,783 that you have permission problems. 901 00:26:24,950 –> 00:26:27,018 You weren’t aware existed, things 902 00:26:27,018 –> 00:26:28,486 that have been shared over time 903 00:26:28,486 –> 00:26:29,688 that are still open 904 00:26:29,921 –> 00:26:31,723 and now maybe should have been locked down. 905 00:26:32,824 –> 00:26:33,558 So we’re talking 906 00:26:33,558 –> 00:26:34,793 about the 907 00:26:34,793 –> 00:26:36,061 data governance and protection 908 00:26:36,227 –> 00:26:39,097 begins with understanding the data that’s out there. 909 00:26:39,331 –> 00:26:39,531 Right. 910 00:26:39,531 –> 00:26:41,499 What do I have that I’m going to surface 911 00:26:41,499 –> 00:26:42,701 to CoPilot. 912 00:26:42,801 –> 00:26:44,002 So that’s able to surface 913 00:26:44,002 –> 00:26:46,237 that on to your employees. 914 00:26:47,272 –> 00:26:49,140 How do I protect that data. 915 00:26:49,708 –> 00:26:51,977 How am I able to secure that in a way 916 00:26:52,344 –> 00:26:54,613 that they’re not seeing things maybe they shouldn’t 917 00:26:54,713 –> 00:26:56,381 within different parts of the organization. 918 00:26:57,115 –> 00:26:57,682 And finally, 919 00:26:57,682 –> 00:26:58,883 and this isn’t always the most 920 00:26:58,883 –> 00:27:00,085 obvious one is 921 00:27:00,118 –> 00:27:01,586 you want to keep the data you need 922 00:27:01,953 –> 00:27:02,887 and remove the data. 923 00:27:02,887 –> 00:27:04,522 You don’t. Right. 924 00:27:05,156 –> 00:27:06,925 This is important for cleaning this up. 925 00:27:07,192 –> 00:27:08,960 This gets into the quality 926 00:27:08,960 –> 00:27:10,161 of the CoPilot results 927 00:27:10,161 –> 00:27:11,429 that you’re going to get quality 928 00:27:11,429 –> 00:27:12,731 the AI search results 929 00:27:12,731 –> 00:27:13,932 that you’re going to get out of this. 930 00:27:14,466 –> 00:27:14,966 Right. 931 00:27:14,966 –> 00:27:16,801 If I have a bunch of old junk 932 00:27:16,801 –> 00:27:18,103 out there, it’s going to surface 933 00:27:18,103 –> 00:27:19,738 that the same as that surfacing the new 934 00:27:20,171 –> 00:27:21,673 and make inferences off of that, 935 00:27:21,906 –> 00:27:23,842 making inferences that maybe are outdated. 936 00:27:24,209 –> 00:27:25,677 They’re based on bad information. 937 00:27:25,977 –> 00:27:26,978 So it’s important that you’re 938 00:27:26,978 –> 00:27:28,246 cleaning up your environment 939 00:27:28,580 –> 00:27:29,714 and making sure that you’re 940 00:27:29,714 –> 00:27:30,982 only have the information 941 00:27:30,982 –> 00:27:32,183 in there that you should. 942 00:27:32,317 –> 00:27:33,518 First thing here 943 00:27:33,718 –> 00:27:34,919 you got to define a data 944 00:27:35,320 –> 00:27:36,855 data governance and protection policy. 945 00:27:36,855 –> 00:27:38,156 So starts 946 00:27:38,156 –> 00:27:39,758 with sensitive data definitions 947 00:27:40,592 –> 00:27:41,292 right. 948 00:27:41,292 –> 00:27:43,361 What am I defining as different levels 949 00:27:43,361 –> 00:27:45,296 of sensitive data within my organization 950 00:27:46,031 –> 00:27:47,065 with Purview. 951 00:27:47,065 –> 00:27:49,000 Microsoft’s got some great schemes for this 952 00:27:49,000 –> 00:27:51,136 and a great progression for evolving this over time. 953 00:27:51,670 –> 00:27:53,805 This is a big, big thing. 954 00:27:54,305 –> 00:27:56,141 This takes time to get it right. 955 00:27:56,608 –> 00:27:57,809 But you can get started 956 00:27:57,809 –> 00:27:59,177 small and start 957 00:27:59,177 –> 00:28:00,578 to realize some of the benefits 958 00:28:00,779 –> 00:28:02,580 and evolve that practice over time. 959 00:28:02,947 –> 00:28:04,683 You don’t have to solve this all at once, 960 00:28:04,949 –> 00:28:06,351 but you do need to pay attention to it, 961 00:28:07,419 –> 00:28:08,019 right? 962 00:28:08,019 –> 00:28:09,220 The other side of that 963 00:28:09,454 –> 00:28:11,089 is having retention policies. 964 00:28:11,523 –> 00:28:12,490 This is what governs 965 00:28:12,490 –> 00:28:14,292 what’s there and what’s gets removed. 966 00:28:14,459 –> 00:28:15,760 What needs to be kept 967 00:28:15,927 –> 00:28:18,063 persistently archived, 968 00:28:18,530 –> 00:28:20,265 and what needs to be deleted. 969 00:28:20,565 –> 00:28:21,766 After time. 970 00:28:22,867 –> 00:28:23,768 I will tell you, 971 00:28:23,768 –> 00:28:24,969 this has been something that, 972 00:28:25,804 –> 00:28:27,272 legal counsel 973 00:28:27,272 –> 00:28:28,506 within certain organizations 974 00:28:28,506 –> 00:28:30,575 have been driving for for some time, 975 00:28:31,042 –> 00:28:31,976 because the more data 976 00:28:31,976 –> 00:28:32,877 that’s in there, 977 00:28:32,877 –> 00:28:35,113 the more data is ultimately discoverable, while what’s 978 00:28:35,280 –> 00:28:37,115 what’s been happening there becomes 979 00:28:37,115 –> 00:28:39,484 any legal challenge, a bigger issue, 980 00:28:39,617 –> 00:28:42,987 more costly, and exposes more risk. 981 00:28:42,987 –> 00:28:44,422 The organization ultimately. 982 00:28:45,590 –> 00:28:46,658 So it’s important again 983 00:28:46,658 –> 00:28:47,859 to clean up the data 984 00:28:48,059 –> 00:28:48,960 and mark the data 985 00:28:48,960 –> 00:28:50,295 so you know what’s sensitive 986 00:28:50,528 –> 00:28:51,730 and what’s not right. 987 00:28:51,730 –> 00:28:52,897 Not everything is sensitive. 988 00:28:53,631 –> 00:28:54,332 Some things 989 00:28:54,332 –> 00:28:55,800 while it’s corporate data, 990 00:28:56,835 –> 00:28:58,036 somebody else saw it. 991 00:28:58,570 –> 00:28:58,970 They’re not 992 00:28:58,970 –> 00:29:00,271 they’re not going to cause any problems. 993 00:29:00,271 –> 00:29:01,439 Right. They know that already. 994 00:29:02,507 –> 00:29:03,274 There are other things. 995 00:29:03,274 –> 00:29:03,975 Yeah. 996 00:29:03,975 –> 00:29:05,844 Mission time Appliance Institute 997 00:29:06,444 –> 00:29:07,178 and stuff like that. 998 00:29:07,178 –> 00:29:08,313 Which you have to retain. 999 00:29:08,313 –> 00:29:10,048 Yeah. There are there are. So yeah. 1000 00:29:10,048 –> 00:29:11,516 So when you’re designing this particularly 1001 00:29:11,516 –> 00:29:12,717 the retention policies 1002 00:29:13,084 –> 00:29:14,018 that when you’re going to make sure 1003 00:29:14,018 –> 00:29:15,754 that to have legal counsel as part of this. 1004 00:29:16,588 –> 00:29:16,988 Absolutely. 1005 00:29:16,988 –> 00:29:18,223 A lot of organizations 1006 00:29:18,223 –> 00:29:20,024 will have their retention 1007 00:29:20,024 –> 00:29:21,760 policies, their archiving policies. 1008 00:29:21,893 –> 00:29:23,862 There’s certain data and depends on your industry. 1009 00:29:23,862 –> 00:29:25,230 That’s not true of one company. 1010 00:29:25,230 –> 00:29:26,898 One size does not fit all here. 1011 00:29:27,398 –> 00:29:28,199 Some companies 1012 00:29:28,199 –> 00:29:29,934 need to keep data forever. 1013 00:29:30,201 –> 00:29:31,736 Some. And it’s certain types of data. 1014 00:29:31,736 –> 00:29:33,037 It’s not everything right. 1015 00:29:33,371 –> 00:29:35,473 Some need to keep it for seven years. 1016 00:29:35,807 –> 00:29:36,775 Some there’s different 1017 00:29:36,775 –> 00:29:38,610 regulatory timelines on all that. 1018 00:29:38,610 –> 00:29:40,044 So it’s important 1019 00:29:40,044 –> 00:29:41,579 we’re not going to drill through all this today. 1020 00:29:41,579 –> 00:29:42,747 We can help with that. 1021 00:29:42,747 –> 00:29:44,048 But we’re not going to do all that today. 1022 00:29:44,082 –> 00:29:45,283 So it’s important to be approaching 1023 00:29:45,850 –> 00:29:47,218 things that things to be considered. 1024 00:29:49,354 –> 00:29:50,555 Next part of this is 1025 00:29:50,822 –> 00:29:52,423 if I’m rolling out CoPilot, 1026 00:29:52,824 –> 00:29:54,526 I need to have an AI enablement plan. 1027 00:29:55,160 –> 00:29:56,060 How am I going to make sure 1028 00:29:56,060 –> 00:29:57,428 everyone’s getting the most value 1029 00:29:57,428 –> 00:29:59,130 for what you’re paying for? Right? 1030 00:29:59,230 –> 00:30:01,366 What am I doing to help drive usage, 1031 00:30:01,733 –> 00:30:02,934 to help drive 1032 00:30:03,067 –> 00:30:05,236 the the realization of the benefits 1033 00:30:05,236 –> 00:30:06,671 that we’re hoping to get from that 1034 00:30:07,839 –> 00:30:09,274 training, ongoing assistance 1035 00:30:09,274 –> 00:30:10,875 they’re doing, 1036 00:30:11,075 –> 00:30:13,278 basic CoPilot, agent creation. 1037 00:30:13,278 –> 00:30:14,512 If you decide that we’re going to start 1038 00:30:14,879 –> 00:30:16,281 allowing for agent creation 1039 00:30:16,281 –> 00:30:17,816 inside of their writing guidance 1040 00:30:17,816 –> 00:30:19,117 for their what’s allowed, what’s not, 1041 00:30:20,485 –> 00:30:22,220 and ensuring that the data sources 1042 00:30:22,220 –> 00:30:23,621 made available to CoPilot 1043 00:30:23,888 –> 00:30:25,089 are the ones you want. 1044 00:30:25,456 –> 00:30:26,691 Right? Could be SharePoint. 1045 00:30:26,691 –> 00:30:28,793 You’d say, hey, we’ll let you operate off 1046 00:30:28,793 –> 00:30:30,428 of these select SharePoint sites, 1047 00:30:30,428 –> 00:30:32,063 but not these, right? 1048 00:30:32,363 –> 00:30:34,132 Allow for OneDrive teams. 1049 00:30:35,700 –> 00:30:36,301 Could be like, 1050 00:30:36,301 –> 00:30:38,036 hey, we’re going to allow CoPilot to interact 1051 00:30:38,036 –> 00:30:40,672 with our Dynamics 365 environment. 1052 00:30:40,939 –> 00:30:42,941 Could be some other external data source. 1053 00:30:43,208 –> 00:30:44,709 You could expose a lot through there. 1054 00:30:45,109 –> 00:30:46,945 But you want to be mindful of what you’re allowing for 1055 00:30:46,945 –> 00:30:48,947 and what you’re not going to just turn on everything. 1056 00:30:50,181 –> 00:30:51,082 Finally, continue 1057 00:30:51,082 –> 00:30:52,550 up to your AI policy as you’re 1058 00:30:52,550 –> 00:30:53,751 making these decisions. 1059 00:30:53,918 –> 00:30:55,086 Make sure that’s captured 1060 00:30:55,086 –> 00:30:56,421 within your AI policy 1061 00:30:57,922 –> 00:30:59,123 education. 1062 00:30:59,357 –> 00:31:00,291 If you’re going to pursue 1063 00:31:00,291 –> 00:31:01,492 data protection 1064 00:31:01,693 –> 00:31:02,727 and retention, 1065 00:31:02,727 –> 00:31:03,995 need to educate people on it. 1066 00:31:04,629 –> 00:31:05,263 You can tag 1067 00:31:05,263 –> 00:31:06,731 all you want, but you need to tell people 1068 00:31:06,731 –> 00:31:08,233 what’s what because they’re going to have 1069 00:31:08,233 –> 00:31:09,667 to help you with the tagging and tuning 1070 00:31:09,667 –> 00:31:10,869 that over time. 1071 00:31:11,469 –> 00:31:12,537 CoPilot use. 1072 00:31:12,537 –> 00:31:14,272 So a host of online training for this. 1073 00:31:14,272 –> 00:31:15,473 A lot of great stuff out there. 1074 00:31:15,940 –> 00:31:17,208 Again, something we can help with. 1075 00:31:18,009 –> 00:31:18,743 But you want to make sure 1076 00:31:18,743 –> 00:31:19,944 that people understand 1077 00:31:20,211 –> 00:31:21,412 how to use the tool 1078 00:31:21,546 –> 00:31:22,580 and get them excited about it, 1079 00:31:22,580 –> 00:31:24,415 get them using it, but doing it the right way. 1080 00:31:25,049 –> 00:31:26,317 And this is about enablement. 1081 00:31:26,317 –> 00:31:28,786 At the end of the day, but doing it in a managed fashion. 1082 00:31:29,888 –> 00:31:31,389 You also may want to consider creating 1083 00:31:31,389 –> 00:31:32,824 a center of excellence, community 1084 00:31:33,157 –> 00:31:34,359 of some kind, 1085 00:31:35,260 –> 00:31:36,728 people that are champions 1086 00:31:36,728 –> 00:31:38,730 for AI, that understand it maybe a little better 1087 00:31:38,730 –> 00:31:39,931 than everybody else, 1088 00:31:39,931 –> 00:31:41,432 and kind of help the organization 1089 00:31:42,100 –> 00:31:43,268 find its way forward. 1090 00:31:43,902 –> 00:31:44,335 Right. 1091 00:31:44,335 –> 00:31:45,536 It’s a great practice. 1092 00:31:45,770 –> 00:31:47,839 Can be hard to execute, 1093 00:31:48,373 –> 00:31:49,607 but it is a great practice 1094 00:31:49,607 –> 00:31:50,808 when done right. 1095 00:31:50,842 –> 00:31:52,677 Yes. Mean if you can. 1096 00:31:52,677 –> 00:31:54,178 Have a pilot integrated 1097 00:31:55,480 –> 00:31:56,681 that in 1098 00:31:56,781 –> 00:31:58,549 just to help you build your own. 1099 00:31:59,050 –> 00:32:00,218 Work. Yep. 1100 00:32:00,285 –> 00:32:02,587 It’s 100% safe 1101 00:32:02,887 –> 00:32:04,489 from other people searching. 1102 00:32:04,923 –> 00:32:05,990 You know, if you did not 1103 00:32:05,990 –> 00:32:07,191 put it in the service 1104 00:32:07,392 –> 00:32:08,593 and you’re just using 1105 00:32:08,960 –> 00:32:11,062 100% saved, that 1106 00:32:11,296 –> 00:32:12,563 when someone does a search 1107 00:32:12,764 –> 00:32:15,099 like Daryl and Daryl did. 1108 00:32:16,234 –> 00:32:18,303 So I’m going to categorize 1109 00:32:18,303 –> 00:32:19,771 that answer a little bit classified, but 1110 00:32:20,271 –> 00:32:22,206 so M365 CoPilot, 1111 00:32:22,206 –> 00:32:23,408 the one you paid for, 1112 00:32:23,474 –> 00:32:24,676 you have your session. 1113 00:32:25,009 –> 00:32:25,810 That information 1114 00:32:25,810 –> 00:32:27,011 is contained within the session. 1115 00:32:27,211 –> 00:32:29,113 The other thing, through Certified CoPilot does 1116 00:32:29,414 –> 00:32:30,782 is it runs off the permissions 1117 00:32:30,782 –> 00:32:31,983 within the Microsoft platform. 1118 00:32:32,150 –> 00:32:33,484 So if you can’t see it, 1119 00:32:33,985 –> 00:32:36,087 if you’re not able to find it and search anyways 1120 00:32:36,120 –> 00:32:37,388 somebody else’s OneDrive 1121 00:32:37,956 –> 00:32:38,823 SharePoint site 1122 00:32:38,823 –> 00:32:40,024 that you don’t have permissions for, 1123 00:32:40,358 –> 00:32:41,926 it will restrict that for sure. 1124 00:32:42,460 –> 00:32:43,127 There’s that is 1125 00:32:43,127 –> 00:32:44,429 core to what that’s providing. 1126 00:32:44,429 –> 00:32:45,697 That’s why it’s such a great tool 1127 00:32:46,130 –> 00:32:47,598 because it handles that for you. 1128 00:32:47,598 –> 00:32:48,933 So you don’t have to to manage 1129 00:32:48,933 –> 00:32:50,134 that the same way. 1130 00:32:50,134 –> 00:32:51,336 Right. 1131 00:32:51,336 –> 00:32:53,304 Your data also is not shared outside 1132 00:32:53,304 –> 00:32:54,539 of the organization at all. 1133 00:32:55,039 –> 00:32:55,907 It’s designed so 1134 00:32:55,907 –> 00:32:57,108 that your CoPilot instance 1135 00:32:58,109 –> 00:32:59,310 physically, well, 1136 00:32:59,610 –> 00:33:00,645 none of it’s physical, 1137 00:33:00,645 –> 00:33:02,280 but there are boundaries 1138 00:33:02,280 –> 00:33:04,215 to what that CoPilot instance can do. 1139 00:33:04,449 –> 00:33:06,250 It’s not tied in with anybody else’s. 1140 00:33:06,484 –> 00:33:08,920 So your data can’t flow outside of that unless you 1141 00:33:09,454 –> 00:33:10,188 actually share that 1142 00:33:10,188 –> 00:33:11,389 with somebody else. 1143 00:33:11,689 –> 00:33:12,857 Okay. 1144 00:33:12,857 –> 00:33:13,257 Great. 1145 00:33:13,257 –> 00:33:14,459 Great question by the way. 1146 00:33:16,127 –> 00:33:17,161 So let’s talk about protection 1147 00:33:17,161 –> 00:33:18,363 monitoring 1148 00:33:18,796 –> 00:33:20,064 this speaking to what 1149 00:33:20,064 –> 00:33:21,265 we were just talking about. 1150 00:33:22,166 –> 00:33:23,868 One of the first things I would do 1151 00:33:23,868 –> 00:33:25,570 is review your SharePoint permissions. 1152 00:33:26,204 –> 00:33:28,940 Anybody with a single CoPilot M365 1153 00:33:28,940 –> 00:33:30,141 CoPilot license 1154 00:33:30,441 –> 00:33:32,477 also has the rights to SharePoint 1155 00:33:32,477 –> 00:33:33,678 Advanced Management. 1156 00:33:33,878 –> 00:33:34,879 It’s a suite of tools 1157 00:33:34,879 –> 00:33:36,080 that Microsoft’s provided 1158 00:33:36,381 –> 00:33:37,448 allows you to review 1159 00:33:37,448 –> 00:33:39,817 permissions, review your top 100 1160 00:33:39,817 –> 00:33:41,019 SharePoint sites to see 1161 00:33:41,019 –> 00:33:42,220 what’s getting hit the most, 1162 00:33:42,787 –> 00:33:43,888 and be able to govern 1163 00:33:43,888 –> 00:33:45,656 what’s shared within CoPilot. 1164 00:33:46,791 –> 00:33:48,226 Very, very useful. 1165 00:33:48,493 –> 00:33:50,728 It’s eye opening for most organizations. 1166 00:33:50,728 –> 00:33:52,230 Everyone thinks they got it locked down. 1167 00:33:52,663 –> 00:33:54,465 You’d be surprised what’s happened over time. 1168 00:33:54,599 –> 00:33:55,933 I’ll be honest. 1169 00:33:56,634 –> 00:33:57,235 So that’s 1170 00:33:57,235 –> 00:33:58,536 that’s the first place to start. 1171 00:33:58,770 –> 00:33:59,337 Permissions. 1172 00:33:59,337 –> 00:34:01,239 Start stops, new access. 1173 00:34:01,739 –> 00:34:03,608 What it doesn’t stop is people 1174 00:34:03,608 –> 00:34:04,809 that have access 1175 00:34:05,009 –> 00:34:05,910 may be sharing that 1176 00:34:05,910 –> 00:34:07,245 in different ways that they shouldn’t. 1177 00:34:07,945 –> 00:34:09,447 That’s where sensitive data 1178 00:34:09,981 –> 00:34:11,816 management and Purview comes in. 1179 00:34:12,250 –> 00:34:13,251 That then takes that down 1180 00:34:13,251 –> 00:34:14,452 to the document level. 1181 00:34:14,886 –> 00:34:16,320 If I have access to something, 1182 00:34:17,055 –> 00:34:18,156 I say, oh, that’s great. 1183 00:34:19,323 –> 00:34:19,690 My friend 1184 00:34:19,690 –> 00:34:20,892 Levi back there was like, hey, 1185 00:34:21,292 –> 00:34:22,293 can you share that with me? 1186 00:34:22,293 –> 00:34:23,795 Oh sure. Not thinking about it. 1187 00:34:24,195 –> 00:34:24,762 Here you go. 1188 00:34:24,762 –> 00:34:25,963 But he shouldn’t have that. 1189 00:34:26,664 –> 00:34:28,166 I just transferring that document over. 1190 00:34:28,166 –> 00:34:29,367 Nothing stops that. 1191 00:34:29,767 –> 00:34:31,736 If it’s a sensitive documents, 1192 00:34:31,736 –> 00:34:33,171 you want to be able to protect 1193 00:34:33,171 –> 00:34:34,372 that at the document level. 1194 00:34:34,539 –> 00:34:36,174 So he goes to get that file, 1195 00:34:36,574 –> 00:34:37,775 he can’t open it. 1196 00:34:37,909 –> 00:34:39,310 The same thing will be respected 1197 00:34:39,310 –> 00:34:40,812 within CoPilot will say 1198 00:34:40,845 –> 00:34:42,613 you’re not going to be able you don’t have rights to this. 1199 00:34:42,613 –> 00:34:43,915 I’m not going to infer things off 1200 00:34:43,915 –> 00:34:45,116 of that information. 1201 00:34:45,149 –> 00:34:47,418 So provides that underlying protection 1202 00:34:48,086 –> 00:34:49,020 for those documents, 1203 00:34:49,020 –> 00:34:50,588 even if it’s someplace it shouldn’t be. 1204 00:34:50,888 –> 00:34:52,290 So it’s that next level down 1205 00:34:52,290 –> 00:34:53,491 from controlling permissions 1206 00:34:54,659 –> 00:34:55,626 retention policies. 1207 00:34:55,626 –> 00:34:56,794 Again cleaning up the data. 1208 00:34:56,794 –> 00:34:58,329 So you’re getting good. 1209 00:34:58,329 –> 00:34:59,697 It’s doing analysis 1210 00:34:59,697 –> 00:35:02,033 and inferences on a quality set of data 1211 00:35:02,633 –> 00:35:04,869 a clean set of data not data 1212 00:35:04,869 –> 00:35:07,839 that’s old outdated unimportant. 1213 00:35:09,540 –> 00:35:09,907 Make sure you 1214 00:35:09,907 –> 00:35:12,043 refine your access policies for CoPilot. 1215 00:35:12,410 –> 00:35:13,611 With entering Intune 1216 00:35:14,045 –> 00:35:14,679 you want to make sure you’re 1217 00:35:14,679 –> 00:35:17,648 driving MFA in the right way. 1218 00:35:17,648 –> 00:35:19,150 So if somebody is accessing 1219 00:35:19,484 –> 00:35:21,786 CoPilot, it’s just like drilling 1220 00:35:21,786 –> 00:35:23,287 right into SharePoint anywhere they are. 1221 00:35:23,588 –> 00:35:24,856 You want to make sure you’re protecting 1222 00:35:24,856 –> 00:35:27,425 CoPilot like you would any other way 1223 00:35:27,425 –> 00:35:29,260 of accessing your corporate information. 1224 00:35:30,394 –> 00:35:31,629 And then I would start monitoring 1225 00:35:31,629 –> 00:35:32,830 the CoPilot prompts 1226 00:35:33,064 –> 00:35:33,698 Purview again 1227 00:35:33,698 –> 00:35:34,999 has a suite of tools here. 1228 00:35:35,233 –> 00:35:36,434 Communication 1229 00:35:36,434 –> 00:35:37,635 Compliance is a great one. 1230 00:35:37,969 –> 00:35:39,170 I can start monitoring 1231 00:35:39,370 –> 00:35:40,638 the prompts people are putting 1232 00:35:40,638 –> 00:35:42,273 through and start realizing, 1233 00:35:42,273 –> 00:35:43,875 hey, we’re not meeting this need 1234 00:35:43,875 –> 00:35:46,511 or we need to put a stop to this, right? 1235 00:35:46,511 –> 00:35:47,812 You gain a lot of information 1236 00:35:47,812 –> 00:35:50,781 from that insider risk auditing. 1237 00:35:51,582 –> 00:35:52,450 And again, 1238 00:35:52,450 –> 00:35:53,718 data security, posture management 1239 00:35:54,051 –> 00:35:55,019 and then eDiscovery. 1240 00:35:55,019 –> 00:35:56,854 eDiscovery allows you to search all this too. 1241 00:35:57,121 –> 00:35:59,023 There are some things you’re going to find in CoPilot. 1242 00:35:59,290 –> 00:36:00,458 You have to put on hold. 1243 00:36:01,125 –> 00:36:01,359 Right. 1244 00:36:01,359 –> 00:36:03,060 That’s now information exchange 1245 00:36:03,361 –> 00:36:04,929 that has to be captured there too. 1246 00:36:05,229 –> 00:36:06,831 So it’s important to have these tools there 1247 00:36:06,831 –> 00:36:07,965 so you can actually capture that. 1248 00:36:10,434 –> 00:36:11,636 Testing. 1249 00:36:11,669 –> 00:36:12,537 So now since we’re 1250 00:36:12,537 –> 00:36:14,172 actually using AI and using it 1251 00:36:14,438 –> 00:36:15,973 internally in a managed fashion, 1252 00:36:16,707 –> 00:36:17,675 you need to be testing 1253 00:36:17,675 –> 00:36:18,876 this periodically. 1254 00:36:19,043 –> 00:36:21,345 Running test prompts, running M365 1255 00:36:21,345 –> 00:36:23,014 search understand 1256 00:36:23,948 –> 00:36:25,750 M365 graph and search 1257 00:36:25,750 –> 00:36:27,585 data is the foundation 1258 00:36:27,585 –> 00:36:30,421 that’s underneath M365 CoPilot. 1259 00:36:31,088 –> 00:36:32,290 If I can search for it, 1260 00:36:32,590 –> 00:36:33,991 if I can prompt for it, 1261 00:36:34,358 –> 00:36:36,394 it’s all the same, right? 1262 00:36:36,761 –> 00:36:39,163 So I think the interesting part there is that 1263 00:36:39,163 –> 00:36:42,166 you actually can start bringing in other data sources 1264 00:36:42,700 –> 00:36:43,901 just by incorporating 1265 00:36:43,901 –> 00:36:45,236 that into the Microsoft 1266 00:36:45,236 –> 00:36:46,671 365 search fabric, 1267 00:36:47,138 –> 00:36:48,306 so I can start bringing on 1268 00:36:48,306 –> 00:36:49,507 premise data sources 1269 00:36:49,774 –> 00:36:51,842 and start to be able to surface insights 1270 00:36:51,876 –> 00:36:53,911 off of that because of this fact. 1271 00:36:54,445 –> 00:36:56,380 So it’s an important aspect of what’s happening 1272 00:36:56,380 –> 00:36:57,582 behind the scenes. Are. 1273 00:36:59,784 –> 00:37:02,119 Now, as we’re going past Productized 1274 00:37:02,119 –> 00:37:04,188 AI and starting to develop maybe 1275 00:37:04,222 –> 00:37:05,957 agents building our own 1276 00:37:06,991 –> 00:37:08,192 shared responsibility model 1277 00:37:08,192 –> 00:37:09,393 starts coming into play. 1278 00:37:10,628 –> 00:37:11,829 Just using CoPilot. 1279 00:37:12,029 –> 00:37:13,164 I’m managing for usage. 1280 00:37:13,164 –> 00:37:15,399 That’s what we’ve been talking about so far, right? 1281 00:37:15,967 –> 00:37:17,168 I start building, 1282 00:37:17,468 –> 00:37:19,170 I have further responsibilities. 1283 00:37:19,170 –> 00:37:20,571 I’m now offering a product 1284 00:37:21,072 –> 00:37:22,640 to my own internal customers, 1285 00:37:22,840 –> 00:37:24,508 possibly even external customers, 1286 00:37:24,976 –> 00:37:25,776 that I need to start 1287 00:37:25,776 –> 00:37:27,044 managing things differently. 1288 00:37:27,278 –> 00:37:29,113 Start managing the underlying model. 1289 00:37:29,380 –> 00:37:29,647 Right. 1290 00:37:29,647 –> 00:37:30,982 You’re not building CoPilot, 1291 00:37:31,315 –> 00:37:32,783 you’re building some agents on top of that. 1292 00:37:33,017 –> 00:37:34,518 But if I’m starting to take that further 1293 00:37:34,518 –> 00:37:36,520 and developing my own models on top of there, 1294 00:37:36,954 –> 00:37:37,989 I have some responsibility 1295 00:37:37,989 –> 00:37:39,190 to manage that the right way. 1296 00:37:41,659 –> 00:37:43,594 So let’s take the last final stage. 1297 00:37:43,894 –> 00:37:46,330 Starting to build AI offerings here. 1298 00:37:46,330 –> 00:37:47,865 Again, you have to assume responsibility 1299 00:37:47,865 –> 00:37:49,066 for what you’re building. 1300 00:37:49,500 –> 00:37:50,501 You also need to protect 1301 00:37:50,501 –> 00:37:51,702 your organization 1302 00:37:51,702 –> 00:37:52,837 against unexpected 1303 00:37:52,837 –> 00:37:54,639 or misleading information sharing. 1304 00:37:54,905 –> 00:37:56,474 This is probably a little different at this point, 1305 00:37:56,474 –> 00:37:57,675 because I know I have 1306 00:37:57,675 –> 00:37:59,110 a responsibility to provide 1307 00:37:59,343 –> 00:38:00,645 a trusted tool 1308 00:38:00,645 –> 00:38:01,879 to anybody that’s using it. 1309 00:38:02,313 –> 00:38:04,448 I need to be managing that now far differently 1310 00:38:04,448 –> 00:38:06,117 than I used to with CoPilot. 1311 00:38:06,117 –> 00:38:08,185 A lot of that’s handled within the CoPilot fabric. 1312 00:38:08,519 –> 00:38:10,655 If I’m going to do this myself, I need to make sure that I’m 1313 00:38:10,655 –> 00:38:11,922 providing quality results 1314 00:38:12,323 –> 00:38:13,724 in order to have that be useful 1315 00:38:13,991 –> 00:38:15,593 and to realize, again, the benefits. 1316 00:38:15,893 –> 00:38:17,328 Ultimately, I’m expecting off of that. 1317 00:38:20,431 –> 00:38:20,965 Let’s see what these 1318 00:38:20,965 –> 00:38:22,166 guys are up to now. 1319 00:38:23,901 –> 00:38:24,635 Hey, Darryl. 1320 00:38:24,635 –> 00:38:25,936 Yeah, this is the new, 1321 00:38:25,936 –> 00:38:27,338 air agent that they just released. 1322 00:38:27,438 –> 00:38:29,140 Yeah, he spent a lot of time working on that one. 1323 00:38:29,240 –> 00:38:31,108 Yeah, it’s really cool, a minor unveil. 1324 00:38:31,142 –> 00:38:32,243 Yeah. Here. 1325 00:38:32,243 –> 00:38:32,710 His deal. 1326 00:38:32,710 –> 00:38:35,513 So, basically, this answers to different questions 1327 00:38:35,513 –> 00:38:37,648 about policies that we have at the company. 1328 00:38:37,648 –> 00:38:39,350 So I want to read, like, a really big manual. 1329 00:38:39,383 –> 00:38:40,217 Oh, sweet. 1330 00:38:40,217 –> 00:38:42,586 So let me, I always forget 1331 00:38:42,586 –> 00:38:44,455 how many vacation days. 1332 00:38:44,455 –> 00:38:46,524 Yeah. You know. All right. 1333 00:38:46,524 –> 00:38:48,059 Okay, I might as well just ask 1334 00:38:48,059 –> 00:38:49,260 the new agent there. 1335 00:38:49,794 –> 00:38:52,330 How many GTO days 1336 00:38:52,330 –> 00:38:54,632 do I get in a given year? 1337 00:38:54,865 –> 00:38:56,801 And this is, like, 100% correct, right? 1338 00:38:56,834 –> 00:38:57,535 Yeah. Yeah. 1339 00:38:57,535 –> 00:38:58,336 Everything comes about. 1340 00:38:58,336 –> 00:38:59,603 Data is all sweet. 1341 00:39:00,304 –> 00:39:01,505 All right. 1342 00:39:01,505 –> 00:39:02,773 Oh, wow. 1343 00:39:03,174 –> 00:39:06,110 You get 80 days of PTO, 80 days off. 1344 00:39:06,143 –> 00:39:07,878 Are you kidding me with this one? 1345 00:39:07,945 –> 00:39:09,580 Darryl. That’s awesome. Wow. 1346 00:39:09,613 –> 00:39:12,049 All right, I need to plan a trip right now. 1347 00:39:12,483 –> 00:39:12,883 Where? 1348 00:39:12,883 –> 00:39:14,185 I’m ready to use this new agent. 1349 00:39:14,251 –> 00:39:15,953 Why does this take, like, a month off? 1350 00:39:16,354 –> 00:39:17,688 I love this agent. All right? 1351 00:39:17,722 –> 00:39:19,457 I mean, I mean, I reached out to my girlfriend, 1352 00:39:19,457 –> 00:39:21,125 actually, she could be so excited about this. 1353 00:39:21,525 –> 00:39:23,094 You know, I’m putting my PTO on right now. 1354 00:39:25,229 –> 00:39:26,630 you said, oh, I love this agent. 1355 00:39:26,630 –> 00:39:28,933 This is great, but it’s not giving you good information. 1356 00:39:28,933 –> 00:39:29,233 Right. 1357 00:39:29,233 –> 00:39:30,468 And they’re about to make some bad 1358 00:39:30,468 –> 00:39:32,303 decisions based on on 1359 00:39:32,303 –> 00:39:33,504 bad information. 1360 00:39:33,504 –> 00:39:35,172 Hence the responsibility as you’re building 1361 00:39:35,172 –> 00:39:37,608 this agent, the company built this agent in this case. 1362 00:39:37,608 –> 00:39:39,377 And they’re theoretical example 1363 00:39:39,810 –> 00:39:41,011 for these gentlemen. 1364 00:39:42,480 –> 00:39:43,414 So as you’re 1365 00:39:43,414 –> 00:39:44,615 diving into building your own 1366 00:39:44,615 –> 00:39:46,083 agent it’s important. 1367 00:39:46,183 –> 00:39:47,385 First of all, 1368 00:39:47,451 –> 00:39:49,453 do you have a broader AI strategy now. 1369 00:39:49,487 –> 00:39:52,390 Now you’re taking the some of further step 1370 00:39:52,390 –> 00:39:53,657 into what you’re doing with 1371 00:39:53,657 –> 00:39:55,559 I need to have a plan. 1372 00:39:55,559 –> 00:39:57,061 You need to have a vision for yourself. 1373 00:39:57,261 –> 00:39:58,596 You need to establish some priorities 1374 00:39:58,596 –> 00:39:59,797 for yourself. 1375 00:39:59,830 –> 00:40:01,232 Need to have a model for 1376 00:40:01,232 –> 00:40:03,300 how do I evaluate different I use cases 1377 00:40:03,601 –> 00:40:04,769 Brian highlighted 1378 00:40:04,769 –> 00:40:05,970 a lot of this this morning. 1379 00:40:06,437 –> 00:40:06,771 Right. 1380 00:40:06,771 –> 00:40:08,406 How do I manage a portfolio of agents. 1381 00:40:08,406 –> 00:40:09,473 What’s that going to look like? 1382 00:40:09,473 –> 00:40:11,041 How am I going to make decisions right. 1383 00:40:11,542 –> 00:40:13,644 Being able to define 1384 00:40:13,677 –> 00:40:15,012 approved technology and data 1385 00:40:15,246 –> 00:40:16,580 or how you’re going to approve 1386 00:40:16,580 –> 00:40:17,882 technology and data over time, 1387 00:40:18,649 –> 00:40:20,418 have a governance plan requirements. 1388 00:40:21,218 –> 00:40:22,920 And this is probably the part 1389 00:40:22,920 –> 00:40:24,288 that that gets forgotten sometimes. 1390 00:40:24,288 –> 00:40:26,223 Here is what do I need to do 1391 00:40:26,223 –> 00:40:28,893 within the organization to drive this 1392 00:40:29,126 –> 00:40:30,161 and be able to 1393 00:40:30,161 –> 00:40:32,596 to perform this the right way, right. 1394 00:40:32,596 –> 00:40:34,231 What are some of the organizational changes 1395 00:40:34,231 –> 00:40:36,834 that have to occur so that I can both 1396 00:40:37,001 –> 00:40:38,636 develop this and use it 1397 00:40:38,769 –> 00:40:39,970 and then manage it? 1398 00:40:40,037 –> 00:40:42,606 Right now, this isn’t just a productize tool. 1399 00:40:42,873 –> 00:40:44,041 I’m creating something. 1400 00:40:44,241 –> 00:40:45,743 It’s going to involves a little extra, 1401 00:40:46,210 –> 00:40:47,678 thought process. There. 1402 00:40:49,213 –> 00:40:50,014 The other thing you should do 1403 00:40:50,014 –> 00:40:52,249 here is plan an AI framework policy. 1404 00:40:52,616 –> 00:40:53,818 This can be challenging. 1405 00:40:55,052 –> 00:40:55,920 This is an area 1406 00:40:55,920 –> 00:40:57,354 that the tooling is evolving 1407 00:40:57,354 –> 00:40:58,556 as we speak. 1408 00:40:58,622 –> 00:40:58,923 Right? 1409 00:40:58,923 –> 00:41:00,090 Right now 1410 00:41:00,157 –> 00:41:02,193 there’s a lot of good things coming into this space 1411 00:41:02,193 –> 00:41:03,961 because the need for it has been realized. 1412 00:41:04,495 –> 00:41:05,463 But that’s 1413 00:41:05,463 –> 00:41:06,230 the major 1414 00:41:06,230 –> 00:41:07,665 vendors are kind of catching up to 1415 00:41:07,665 –> 00:41:09,066 what’s all going to have to happen here. 1416 00:41:09,567 –> 00:41:10,768 A lot of tools out there. 1417 00:41:11,035 –> 00:41:12,203 I expect over time 1418 00:41:12,203 –> 00:41:13,404 that’s going to consolidate. 1419 00:41:13,437 –> 00:41:14,772 You’re going to have a much simpler framework 1420 00:41:14,772 –> 00:41:15,973 for this over time. 1421 00:41:16,006 –> 00:41:18,108 But let’s talk about what what needs to happen in there today. 1422 00:41:18,108 –> 00:41:20,110 We need to manage 1423 00:41:20,110 –> 00:41:21,312 lifecycle management, 1424 00:41:21,512 –> 00:41:22,680 need to manage security, 1425 00:41:22,680 –> 00:41:25,015 organizational policies 1426 00:41:25,015 –> 00:41:27,718 that need to be factored into this cost management. 1427 00:41:28,252 –> 00:41:29,587 AI development can get 1428 00:41:29,587 –> 00:41:30,821 really costly really fast. 1429 00:41:30,821 –> 00:41:32,656 You need to keep an eye on that right? 1430 00:41:33,791 –> 00:41:34,425 Make sure you’re 1431 00:41:34,425 –> 00:41:36,627 defining what models and tooling for everyone to use. 1432 00:41:36,861 –> 00:41:38,262 So again, driving the security 1433 00:41:38,262 –> 00:41:39,597 and operational aspects of that. 1434 00:41:40,030 –> 00:41:41,298 You’re doing this a smart way. 1435 00:41:41,932 –> 00:41:42,833 This is just about driving 1436 00:41:42,833 –> 00:41:44,034 responsible use, 1437 00:41:44,301 –> 00:41:46,537 not trying to stop anything, 1438 00:41:47,037 –> 00:41:50,074 just trying to protect it so that you can keep going down the road. 1439 00:41:50,074 –> 00:41:52,409 You don’t hit a running up in the ditch 1440 00:41:52,476 –> 00:41:53,677 later. 1441 00:41:54,011 –> 00:41:54,845 Again, continue 1442 00:41:54,845 –> 00:41:56,046 to update the AI policy 1443 00:41:57,548 –> 00:41:58,883 education enablement here 1444 00:41:58,883 –> 00:42:00,384 you’re really focusing on, right? 1445 00:42:00,384 –> 00:42:02,119 How do I create this framework 1446 00:42:02,486 –> 00:42:04,388 for how we’re publishing agents? 1447 00:42:04,655 –> 00:42:06,423 I’m creating AI product out there. 1448 00:42:06,724 –> 00:42:07,892 Right. 1449 00:42:07,892 –> 00:42:09,393 Some of the great tools out there 1450 00:42:09,393 –> 00:42:11,028 Microsoft has as a host of these 1451 00:42:11,028 –> 00:42:13,430 depends on the tooling you’re using and kind of them, 1452 00:42:13,531 –> 00:42:16,066 I think you’re going to see agent 365 slowly 1453 00:42:16,367 –> 00:42:18,269 take over this space more and more. 1454 00:42:18,936 –> 00:42:19,670 But right now you have 1455 00:42:19,670 –> 00:42:20,971 the CoPilot control system. 1456 00:42:20,971 –> 00:42:23,607 So if I’m doing things with CoPilot Studio, 1457 00:42:24,341 –> 00:42:26,076 I have kind of a control plane there. 1458 00:42:26,610 –> 00:42:27,311 I foundry, 1459 00:42:27,311 –> 00:42:28,612 if I’m doing things there, maybe 1460 00:42:28,612 –> 00:42:30,614 using some Azure services for 1461 00:42:31,181 –> 00:42:32,650 I can use the control plane there. 1462 00:42:33,183 –> 00:42:35,619 Agent 365 is emerging as kind of that 1463 00:42:35,886 –> 00:42:37,321 that central framework 1464 00:42:37,321 –> 00:42:38,522 for all of this. 1465 00:42:38,522 –> 00:42:39,890 But it’s maturing as we speak. 1466 00:42:39,890 –> 00:42:41,091 This was just announced 1467 00:42:41,392 –> 00:42:42,760 a few months ago in ignite. 1468 00:42:43,127 –> 00:42:45,663 So it’s an, in, in development. 1469 00:42:45,663 –> 00:42:46,530 Right. It’s on its way. 1470 00:42:48,766 –> 00:42:50,968 Organizationally, you want to consider 1471 00:42:50,968 –> 00:42:52,202 creating the AI product group 1472 00:42:52,202 –> 00:42:53,470 now too, right? 1473 00:42:53,537 –> 00:42:55,472 I need to have somebody that’s owning these things. 1474 00:42:55,472 –> 00:42:56,740 That’s going to be responsible for what 1475 00:42:56,740 –> 00:42:58,375 they’re doing, responsible 1476 00:42:58,375 –> 00:42:59,910 for the ongoing management of this, 1477 00:43:00,945 –> 00:43:02,279 and be able to lead and manage 1478 00:43:02,279 –> 00:43:03,480 the development efforts. 1479 00:43:05,716 –> 00:43:06,817 Agent 365. 1480 00:43:06,817 –> 00:43:08,752 Like we said, this is something 1481 00:43:08,752 –> 00:43:09,987 really excited about this. 1482 00:43:09,987 –> 00:43:12,089 I think this is really going to help a lot of organizations 1483 00:43:12,423 –> 00:43:13,357 get past the, 1484 00:43:13,357 –> 00:43:15,225 okay, we’d like to do this, 1485 00:43:15,225 –> 00:43:16,860 but we’re not quite sure how to manage this. 1486 00:43:17,194 –> 00:43:18,862 I think this is going to help a lot of organizations 1487 00:43:18,862 –> 00:43:20,064 do that very quickly. 1488 00:43:21,498 –> 00:43:22,700 Right now. 1489 00:43:23,767 –> 00:43:24,602 Way to do this. 1490 00:43:24,602 –> 00:43:27,404 The first thing you need to do, 1491 00:43:27,871 –> 00:43:30,474 the sign identity to each agent, 1492 00:43:31,408 –> 00:43:32,610 Microsoft Entourage 1493 00:43:32,810 –> 00:43:33,811 agent identities 1494 00:43:33,811 –> 00:43:35,112 allows you to do this. 1495 00:43:36,480 –> 00:43:37,114 I can then 1496 00:43:37,114 –> 00:43:39,783 monitor what that agent is doing, 1497 00:43:40,351 –> 00:43:41,919 who’s able to interface with it, 1498 00:43:42,152 –> 00:43:43,220 who’s not. 1499 00:43:43,220 –> 00:43:45,422 It allows me, that whole host of controls. 1500 00:43:45,556 –> 00:43:47,458 That identity plane is so important. 1501 00:43:47,458 –> 00:43:49,259 And throughout cloud services 1502 00:43:49,259 –> 00:43:50,461 within Microsoft 1503 00:43:50,561 –> 00:43:51,929 is really going to help out here 1504 00:43:51,929 –> 00:43:53,364 too, right? 1505 00:43:53,364 –> 00:43:54,898 Just an ongoing visibility 1506 00:43:54,898 –> 00:43:56,800 as to who’s interacting with this, 1507 00:43:56,967 –> 00:43:58,602 what this thing’s doing, the data 1508 00:43:58,669 –> 00:43:59,870 that’s accessing 1509 00:44:00,037 –> 00:44:01,839 all kinds of great information. 1510 00:44:01,839 –> 00:44:03,040 There. 1511 00:44:03,741 –> 00:44:05,075 You also want to start here. 1512 00:44:05,075 –> 00:44:06,844 You’re going to probably need to manage your structured 1513 00:44:06,844 –> 00:44:08,612 data state more fabric 1514 00:44:08,612 –> 00:44:09,780 certainly core to this. 1515 00:44:10,147 –> 00:44:12,483 Now you heard from senior earlier on on what 1516 00:44:12,483 –> 00:44:13,651 fabric can do for you. 1517 00:44:13,917 –> 00:44:15,786 Purview also has aspects here too 1518 00:44:15,786 –> 00:44:17,821 that you need to to put into place. 1519 00:44:18,422 –> 00:44:19,990 Purview data catalog can be great 1520 00:44:20,324 –> 00:44:21,525 for helping to surface 1521 00:44:21,759 –> 00:44:22,760 the data sources 1522 00:44:22,760 –> 00:44:24,061 that are out there that are trusted, 1523 00:44:24,495 –> 00:44:26,430 particularly if you’re playing endorsements. 1524 00:44:26,430 –> 00:44:28,399 They’re say, hey, use this data. 1525 00:44:28,832 –> 00:44:30,234 We don’t trust that one so much. 1526 00:44:31,035 –> 00:44:31,769 It’ll speed up 1527 00:44:31,769 –> 00:44:33,003 development drastically. 1528 00:44:33,771 –> 00:44:34,672 The other thing you want to make sure is 1529 00:44:34,672 –> 00:44:35,873 you got the right permissions there. 1530 00:44:36,173 –> 00:44:37,741 If I carry an agent that says, hey, 1531 00:44:38,475 –> 00:44:39,543 here’s all our earnings 1532 00:44:39,543 –> 00:44:40,911 data that anybody in 1533 00:44:40,911 –> 00:44:42,112 the organization could use, 1534 00:44:42,513 –> 00:44:44,181 we may not be hiding it. 1535 00:44:44,415 –> 00:44:45,649 We might be right. 1536 00:44:45,649 –> 00:44:47,084 There may be information I don’t want 1537 00:44:47,084 –> 00:44:49,053 everybody seeing within the organization 1538 00:44:49,620 –> 00:44:50,487 if I’ve got the right 1539 00:44:50,487 –> 00:44:51,722 permissions on that data, 1540 00:44:52,322 –> 00:44:54,692 even if I expose it, potentially 1541 00:44:54,925 –> 00:44:56,760 it’ll actually stop itself from being surfaced 1542 00:44:56,760 –> 00:44:58,529 because of what I try. 1543 00:44:58,562 –> 00:45:00,497 And and the identities will control for 1544 00:45:01,965 –> 00:45:02,866 want to play management 1545 00:45:02,866 –> 00:45:04,068 control system. 1546 00:45:04,435 –> 00:45:06,870 Ultimately I think this is going to be agent 365. 1547 00:45:07,271 –> 00:45:09,339 Right now you’re probably using some amount 1548 00:45:09,339 –> 00:45:12,242 of CoPilot control system that found your control plane 1549 00:45:12,242 –> 00:45:14,611 to put the ongoing management 1550 00:45:14,912 –> 00:45:16,080 for the, 1551 00:45:16,080 –> 00:45:17,281 agents that you’re being developed, 1552 00:45:17,581 –> 00:45:18,916 being able to keep tabs on them. 1553 00:45:19,183 –> 00:45:20,451 Again, identity is very important 1554 00:45:20,451 –> 00:45:21,652 for that. 1555 00:45:22,286 –> 00:45:23,587 You want to be defining access 1556 00:45:23,587 –> 00:45:25,189 policies for agents, right? 1557 00:45:25,189 –> 00:45:26,857 So if I made that agent available, 1558 00:45:27,224 –> 00:45:28,859 where can people access this from? 1559 00:45:29,059 –> 00:45:30,661 Remember, as people 1560 00:45:30,661 –> 00:45:32,930 are interfacing with it, they’re extracting data from it. 1561 00:45:33,430 –> 00:45:34,465 Where do I want that data 1562 00:45:34,465 –> 00:45:35,666 to be able to go? 1563 00:45:35,799 –> 00:45:36,366 Right. 1564 00:45:36,366 –> 00:45:37,768 If I have somebody that says, hey, 1565 00:45:38,102 –> 00:45:39,303 I can log into this, 1566 00:45:39,603 –> 00:45:41,338 but if I’m on my home PC, 1567 00:45:42,039 –> 00:45:44,441 I may not want I may not want them to be able to get that, 1568 00:45:44,875 –> 00:45:46,110 or I may need to control 1569 00:45:46,110 –> 00:45:47,745 that on different devices to say, 1570 00:45:47,745 –> 00:45:50,280 hey, you can’t copy information 1571 00:45:50,280 –> 00:45:51,715 out of this to be able 1572 00:45:51,715 –> 00:45:52,916 to share with anybody else 1573 00:45:53,383 –> 00:45:54,685 but the host of controls 1574 00:45:54,685 –> 00:45:56,453 that you want to have in place to protect yourself. 1575 00:45:58,021 –> 00:45:59,456 Managing agent compliance. 1576 00:46:00,224 –> 00:46:02,559 So here we’re starting to say, okay, 1577 00:46:02,559 –> 00:46:04,027 the agent can do certain things. 1578 00:46:04,528 –> 00:46:05,262 What are some things 1579 00:46:05,262 –> 00:46:07,064 we should be surfacing and not right. 1580 00:46:07,264 –> 00:46:08,398 HIPAA data, 1581 00:46:08,398 –> 00:46:10,234 different regulatory data, 1582 00:46:10,567 –> 00:46:12,269 making sure that that’s controlled 1583 00:46:12,603 –> 00:46:15,105 and managed within the underlying framework. 1584 00:46:15,105 –> 00:46:17,040 So you can protect people from making mistakes there. 1585 00:46:17,441 –> 00:46:18,308 Right. 1586 00:46:18,308 –> 00:46:19,076 A lot of tooling there 1587 00:46:19,076 –> 00:46:20,277 to help you with that. 1588 00:46:21,145 –> 00:46:22,513 Finally, you need to be able to secure 1589 00:46:22,513 –> 00:46:23,881 agents, make sure 1590 00:46:23,881 –> 00:46:25,382 not allowing for unnecessary cyber 1591 00:46:25,382 –> 00:46:26,583 threats there 1592 00:46:26,850 –> 00:46:27,551 and not making 1593 00:46:27,551 –> 00:46:28,752 sure that you’re controlling 1594 00:46:28,786 –> 00:46:30,187 the broader security posture 1595 00:46:30,187 –> 00:46:31,388 management there. 1596 00:46:31,488 –> 00:46:32,256 And also make sure 1597 00:46:32,256 –> 00:46:33,524 that you’re not exposing yourself. 1598 00:46:33,524 –> 00:46:34,725 Some ways you don’t. 1599 00:46:34,725 –> 00:46:35,726 So a combination 1600 00:46:35,726 –> 00:46:37,961 of security posture management 1601 00:46:38,529 –> 00:46:40,531 through defender for cloud right. 1602 00:46:40,531 –> 00:46:42,733 So we’ve got data security posture management 1603 00:46:43,066 –> 00:46:44,234 within Purview 1604 00:46:44,234 –> 00:46:45,435 within defender for cloud. 1605 00:46:45,435 –> 00:46:46,637 There’s 1606 00:46:46,637 –> 00:46:48,038 AI security posture management 1607 00:46:48,038 –> 00:46:49,273 to help control the underlying 1608 00:46:49,273 –> 00:46:50,474 security threats. 1609 00:46:50,641 –> 00:46:52,009 Being able to, to 1610 00:46:52,009 –> 00:46:54,311 allow for injections and all kinds of things. 1611 00:46:54,545 –> 00:46:56,046 It will protect your agents from that. 1612 00:46:57,414 –> 00:46:57,581 You can 1613 00:46:57,581 –> 00:46:59,216 use foundry security, baseline 1614 00:47:00,083 –> 00:47:01,185 red teaming. 1615 00:47:01,185 –> 00:47:02,586 So a host of other tools there too 1616 00:47:02,586 –> 00:47:05,522 that can help with determining 1617 00:47:05,656 –> 00:47:06,990 do I have threats here that maybe 1618 00:47:06,990 –> 00:47:08,192 I’m not aware. 1619 00:47:09,760 –> 00:47:10,961 Testing and development 1620 00:47:11,762 –> 00:47:12,830 certainly want to be monitoring 1621 00:47:12,830 –> 00:47:14,031 the audit logs. 1622 00:47:14,131 –> 00:47:16,099 You need to establish some kind of incident management 1623 00:47:16,099 –> 00:47:17,634 support process around the tooling 1624 00:47:17,634 –> 00:47:18,836 that you’re providing, right. 1625 00:47:18,869 –> 00:47:20,070 You’re providing a product 1626 00:47:20,204 –> 00:47:21,405 you need to be able to support. 1627 00:47:21,605 –> 00:47:23,073 Well, that’s inside the organization 1628 00:47:23,473 –> 00:47:24,675 or outside. 1629 00:47:24,675 –> 00:47:25,876 How am I going to manage that? 1630 00:47:27,211 –> 00:47:28,245 Establish a DevOps 1631 00:47:28,245 –> 00:47:29,546 lifecycle around this right. 1632 00:47:29,813 –> 00:47:31,481 This isn’t going to be a all right. 1633 00:47:31,481 –> 00:47:32,749 It’s all done one time. 1634 00:47:33,050 –> 00:47:34,751 This is going to evolve over time. 1635 00:47:35,519 –> 00:47:36,720 So how am I putting some 1636 00:47:36,720 –> 00:47:38,088 lifecycle management on this. 1637 00:47:38,088 –> 00:47:39,656 So I’m continuing to update 1638 00:47:39,990 –> 00:47:42,092 and make those changes and evolve this as you go. 1639 00:47:43,794 –> 00:47:45,229 As part of that ongoing testing. 1640 00:47:45,229 –> 00:47:46,430 Monitoring is important. 1641 00:47:46,730 –> 00:47:48,932 Having people just go at it, testing 1642 00:47:48,932 –> 00:47:50,701 tools, all kinds of things out there. 1643 00:47:51,802 –> 00:47:52,202 Founder 1644 00:47:52,202 –> 00:47:54,271 and CoPilot have a host of tools built in there 1645 00:47:54,271 –> 00:47:55,472 to help you with this. 1646 00:47:55,672 –> 00:47:57,074 There’s the Microsoft Responsible 1647 00:47:57,074 –> 00:47:58,275 AI toolkit. 1648 00:47:58,508 –> 00:47:59,710 There’s a host of tools 1649 00:47:59,710 –> 00:48:01,345 inside of there that you can leverage towards this, 1650 00:48:01,612 –> 00:48:03,614 something they use themselves for their own internal 1651 00:48:03,847 –> 00:48:05,983 AI development that they make available to you. 1652 00:48:06,550 –> 00:48:07,451 Fair learn helps 1653 00:48:07,451 –> 00:48:09,186 against controlling for biases 1654 00:48:09,620 –> 00:48:11,255 within the the AI models 1655 00:48:11,255 –> 00:48:12,723 and products that you’re developing 1656 00:48:13,857 –> 00:48:15,125 and interpret ML, 1657 00:48:15,125 –> 00:48:16,326 and it kind of ML 1658 00:48:16,360 –> 00:48:18,262 will help with when you’re dealing more with machine 1659 00:48:18,262 –> 00:48:20,097 learning models and doing deeper 1660 00:48:20,097 –> 00:48:21,298 data analysis. 1661 00:48:21,398 –> 00:48:22,699 And make sure that that’s done, 1662 00:48:22,699 –> 00:48:23,901 the right way. 1663 00:48:26,069 –> 00:48:26,503 All right. 1664 00:48:28,238 –> 00:48:28,639 This is 1665 00:48:28,639 –> 00:48:30,307 the spectrum of what we’ve talked about today. 1666 00:48:31,041 –> 00:48:33,710 The purpose of this has been really, 1667 00:48:33,744 –> 00:48:35,579 you know, make sure that you’re anticipating 1668 00:48:36,046 –> 00:48:37,648 what you’re going to be dealing with next. 1669 00:48:37,981 –> 00:48:39,650 It’s not about opposing 1670 00:48:40,083 –> 00:48:42,419 the coming waves of AI development 1671 00:48:42,419 –> 00:48:43,620 that’s going to happen. 1672 00:48:43,620 –> 00:48:45,756 It’s how do you enable that but enable 1673 00:48:45,756 –> 00:48:47,491 that in a responsible way through governance. 1674 00:48:49,660 –> 00:48:50,560 So this is kind of 1675 00:48:50,560 –> 00:48:51,828 a summary of everything. 1676 00:48:51,828 –> 00:48:53,497 We’ve talked about different categories. 1677 00:48:54,131 –> 00:48:55,766 Hopefully this was beneficial to you.