Loading summary
Transcript37 lines
- [00:00]
Tremfiab Representative
Are your ulcerative colitis symptoms proving difficult to manage? Tremfiab can help you manage the cycle of UC symptoms. At one year, many patients taking Tremphya achieved clinical remission and some patients also achieved endoscopic remission. Individual results may vary. Tremphya is a prescription medicine used to treat adults with moderately to severely active ulcerative colitis. Serious allergic reactions and increased risk of infections may occur before treatment. Your doctor should check you for infections and tb. Tell your doctor if you have an infection, flu like symptoms or if you need a vaccine. Ask your doctor if Tremphya can help you manage the cycle of UC symptoms. Call 1-800-526-7736 to learn more or visit tremphyaradio.
- [00:55]
Alan Chappelle
Welcome to the Monopoly Report. The Monopoly Report is dedicated to chronicling and analyzing the impact of antitrust and other regulations on the global advertising economy. If you are new to the Monopoly Report, you can subscribe to our weekly newsletter at Monopoly Market, and you can check out all of the Monopoly report podcasts@monopolyreportpod.com I'm Alan Chappelle. This week my guest is Omer Tenney, a partner in data protection in the law firm of Goodwin Proctor. Omer has been very active within the Future of Privacy Forum and the International association of Privacy Professionals, and I probably known him for 15 years. We even shared a client briefly way back in the day. So I'm excited to get Omer's thoughts on AI, particularly as it ties into a digital ad setting. So let's have at it. Hey Omar, thanks for coming on the pod. How are you?
- [01:48]
Omer Tenney
I'm great. Thanks for inviting Alan. A pleasure being here.
- [01:52]
Alan Chappelle
So today we're mostly going to be talking about, you know, the impact of artificial intelligence, and mostly we talk here about the ad space in particular. But I know your law practice focuses on a whole bunch of different things and so we may veer off of that path a little bit, but everybody in the ad space over the last years is, say, says they're using AI, and in my experience that can mean loads of different things to different people. So my first question to you is like, where does machine learning, which folks in the ad space have been using for two decades, where does that leave off and AI take over? And does that distinction even matter at this point?
- [02:34]
Omer Tenney
There is a definition of an AI system which is pretty common now across different laws in different jurisdictions. And it has, you know, several elements. It has autonomy, which I don't think the machine learning that you mentioned, you know, earlier really did it has inputs and outputs and the outputs can be recommendations or decisions or actually actions, as I said, or content. It has objectives and it can draw inferences. So I think some of those components were there before, but others are clearly, you know, at least reached another level over the past couple of years.
- [03:21]
Alan Chappelle
Yeah, no, fair enough. It's something that I struggle with as I'm talking to clients because it's oftentimes what is being pitched to me sounds very similar to things that were pitched to me five to seven, sometimes even 10 years ago. But I agree with you. The technology, the capabilities are such that it's starting to go well beyond that. So I've sort of taken the position, you know, out in my travels that AI is sort of the new behavioral targeting. And what I mean by that is that so much of the regulatory focus and business interest over the last 15 years has been squarely focused on behavioral targeting. Today all that energy seems to have moved into AI, you know, so what can business and legal folks in the ad space learn from what happened over the last 15 years with BT? You know, any similarities between the two or, or feel free. You can also tell me that I'm full of hot air.
- [04:17]
Omer Tenney
I'm not going to tell you that. And in fact, I think veterans and experts have a lot to teach this new generation of professionals who are focused on AI just because, as you say, behavioral targeting and targeted advertising and the online behavioral advertising space was increasing incredibly central to the discussions of privacy and data protection over the past, I think two decades at least, probably a bit more really since the turn of the millennium, the, the Internet, you know, the dot com bubble in the time when sort of ad supported content became like the reigning business model, the Internet. Right. You know, it's like 85% of the discussion in the privacy and data protection policy legal space revolved around behavioral advertising at the end of the day. Now AI, I think some of the issues are similar, but some are very different. First of all, it's not just about pii. I think for online advertising, PII was really the focus and it was, that was like the oil right of that environment. Whereas with AI there are very different values I feel at play and in fact central to the policy and legal discussion. First and foremost, safety. It's about safety because certainly if you talk about agentic, but even when you just create content, safety issues are front and center. And needless to say, the AI geniuses are kind of warning about the future ADI and artificial general intelligence, which will end humanity and the universe. So that Wasn't sort of on our radar for advertising. Accuracy is really important because we know that these agents hallucinate and they make up stuff, stuff. And you know, that becomes really critical if you use this technology to make impactful decisions about employment, housing, medicine, kind of important things. Bias and discrimination was an issue in ad tech, but I feel it's more central for AI. AI poses risks to intellectual property, to copyright, to patents, raises really thorny and kind of mind bending issues about patentability and copyrightability and what like should or should not be protected. So I think the whole risk surface is very different. And we are now talking, you know, what's the worst thing that could have happened in the behavioral advertising space? You would get like a really creepy kind of targeted ad or something that maybe doesn't fit your interests in profiles, kind of depending on the vantage point. Whereas here the risks are, you know, humanity, democracy, truth. It seems like the stakes are higher.
- [07:48]
Alan Chappelle
Right? But, but you know, we're in the ad space and so the risk to humanity is that really the same? Like I, I'm, I'm sort of struggling with this in my own brain. Is like, is the difference here tangible in terms of overall risks? And like I can think of some areas, particularly in the, in the health space, because as machines start to make odd inferences that are not fully disclosed or even fully aware, then that creates its own set of challenges even as it pertains to the ad space.
- [08:18]
Omer Tenney
No, yeah, look, I think most of the frameworks that are emerging are risk based and risk oriented. And I think behavioral advertising might not be a high risk vector unless you're talking about advertising for, you know, specific products like alone or housing and such, where, you know, issues of, of bias and discrimination could have come up before and they're still present now. So, so I think, you know, maybe this is a change of pace for behavioral advertising experts. Where once you were kind of the use case and the center of attention. And with AI, maybe other sort of industries or you know, areas of the economy are more pertinent, you know.
- [09:18]
Alan Chappelle
You know, it's funny and I've mentioned this in other pods that we've done. In my view, if you somehow magically took precise location and a good deal of the health targeting stuff off the table, somehow I feel like the collective lightning rod that has been the ad space over the last 15 years dissipates pretty significantly.
- [09:41]
Omer Tenney
Look, Alan, it's become almost a political issue with, you know, books written about the surveillance economy. Is it? And in fact the ftc, you know, under the last leadership launching a rulemaking about commercial surveillance. I think, you know, that type of language makes it really a lightning rod and a highly politicized thing because it kind of, it seems like a Snowden issue, but in fact, like, is behavioral advertising sort of as threatening to human dignity, liberty as government surveillance? Probably not.
- [10:30]
Alan Chappelle
Yeah. And look, I have tended to work with smallish venture capital funded ad tech and martech companies more so than like the big tech companies. And I think there's an argument that really never gets made that scale is a huge, huge component to all this. And I think that sort of gets missed in, in some of the surveillance capitalism type arguments. I'm not sure I'm comfortable with that term generally, but as it applies to some tiny ad tech company whose footprint is like a rounding error within the, you know, with, within the universe and just is able to make a couple of, you know, ads slightly more relevant. To me, that maybe isn't as, you know, as world ending, as, as some make it out to be.
- [11:17]
Omer Tenney
It's a complex argument. Let's kind of agree to that because it has many different sort of assets and they cut in different directions. So the competition aspect, you know, power and platforms and kind of how government interacts with it and you know, in some respects, certainly small players are less threatening, but in other ways, you know, they're probably less accountable also and kind of more difficult to nail down and require, you know, transparency and accountability. So I think it's a complex argument, but it's not our topic for du jour.
- [12:03]
Alan Chappelle
Fair point, but, but this sort of builds on my previous question. So my gut sense is, is that there really isn't a regulatory environment out there right now that's really equipped to handle generative AI and LLM. And I'll give you one example, but I don't think it's the only one. But if you were to define personal data in absolute terms, then there's almost no way to fit an LLM into the fair information practice principles. Do you agree? Disagree? Partly agree.
- [12:36]
Omer Tenney
Moving on from basic principles, the question is what do we need to protect from a privacy standpoint in this context of large language models and AI? And I think some things are clear, like we shouldn't let LLMs crush privacy as we've known it, right? If I can ask Chat GPT, like what color underwear does Alan wear? And it'll answer me, then that's a privacy invasion. So we need to bake these things into the technology, right? The algorithm, the machine, and to prevent like the Obvious violations of privacy, but data minimization, sort of trying to prevent the collection of use of information that's primarily publicly available in order to achieve like these scientific breakthroughs. I think you're waging an uphill battle if that's what you're trying to achieve.
- [13:46]
Alan Chappelle
Yeah, it feels like to me that particularly in Europe where I think there's been a tremendous amount of thoughtful analysis around LLMs, that they almost seem to be trying not to say that it's outright illegal. However, if you read through a lot of the documentation and you bring it to its natural conclusion, the only reasonable conclusion is that this stuff is completely illegal. And I don't know how to address that, particularly as maybe it doesn't come to a head for another two years, but by then isn't it too late?
- [14:24]
Omer Tenney
Good point. Look, I think the views in Europe vary even among privacy regulators, but certainly when you bring in kind of other stakeholders. We just heard President Macron on the main stage of this big air conference in Paris last week say that Europe is over regulating the sector. And of course France is trying to advance its own kind of AI unicorn. Mistral in Italy though, guarantee the data protection regulator actually banned chatgpt for a while and I think they might have just done it with Deepseek 2. So I think views range and vary, but banning a technology is not something that's proven successful or worthwhile or beneficial really. And I think the only technology we really ban in certain respects is nuclear. With the nuclear, you know, non proliferation Treaty even there, as you know, a lot of exceptions are popping up and there you kind of need the clout, the resources of a nation state in order to pursue the technology here. If Deep Sea can kind of stand up robust LLM with an investment of like whatever, 6, $7 million, it becomes clear that you're not going to ban it. So try to focus, I think on the harms and as I said, you don't want it to completely eviscerate privacy and to make us all transparent individuals. But you know, let go of kind of old rules that don't seem to apply in this context.
- [16:22]
Alan Chappelle
Nope, that makes sense. So I'm going to shift gears a little bit and increasingly we're seeing laws require some flavor of an AI assessment. I think you're going to see that in a bunch of US state laws. Pretty soon California's working on their version of it with their seemingly endless and unlimited ability to make rules over there. Now walk me through what is an AI assessment, because as I've tried To socialize that concept through, you know, the business community, some with clients of mine, some with just people that I know. You get back kind of a look of like, oh good, that, that's another European thing where those guys just love their paperwork and great, so what do we have to do here to, to walk ourselves through it? So like you're advising companies on this, like what? You know, how do you get them to approach and how do you get buy in to start working on one of those in earnest?
- [17:14]
Omer Tenney
Look, a couple of laws already require these assessments, namely the Colorado AI act, which requires a assessments. And even more sort of specifically and prescriptively, the New York City 144 Law on AI Systems and Employment context requires bias audits before deploying my assistant. In that context, I don't think risk assessments are a European concept at all. And in fact, even in the context that we are kind of thinking about, which is DPIAs, right, data protection, impact assessments, those were born in the U.S. i think the Department of Homeland Security first instituted the PIAs as part of their kind of technology or system adoption cycle. And you know, the NEST AI risk management framework is just that. It's an AI risk management framework which requires performing risk assessment. You know, part of it is bias audits like NYC144 requires. As lawyers, we don't do that stuff, but we, you know, tell clients that they need to if they integrate this into their hiring processes. You know, the law is pretty descriptive actually in terms of what you need to do. You need to kind of measure how different candidates kind of go through the system. And if it ends up demonstrating that, you know, more men get in than women, relatively kind of adjusted for all the variables, then you need to fix it. We're also talking about red teaming, you know, kind of a concept from cyber, but it's been adopted here. And what you do in this respect, you test the system with different prompts, you kind of ask it things that try to make it cross boundaries that it shouldn't, you know, talk about politics or religion or maybe personal issues. And that way kind of tweak the algorithm to make sure that it's safe. But at the end of the day, I think risk assessments are really important. And it goes back to your initial point, like in behavioral advertising, kind of more generally in the advertising context, what's the risk? Well, if you're, you know, if you're advertising hair products and I get like the baldy stuff because that's what I am, you know, you can call it discrimination But I call it like good, sort of useful. Right. Beneficial targeting for my bald pate. Whereas if it's, you know, if it's advertising like different mortgage rates or loans to people based on the color of their skin or religion or, you know, then obviously you should. That's high risk. Right. And you shouldn't tolerate that there are laws against it. And in fact, I like this announcement by the U.S. government a couple of years ago. So just after OpenAI launched ChatGPT, there was an interagency statement by the FTC, CFPB, EEOC, the employment regulator, and the DOJ, April 2023. And they said, we don't have a comprehensive AI law in the United States, but there is also no AI exemption from existing laws. So, you know, if you are using these tools in a way that violates fair housing or kind of employment discrimination laws, then it's a problem.
- [21:13]
Alan Chappelle
Yeah. One of the things that I've done that's worked for me in my practice, and this just actually goes Beyond DPIAs and AI impact assessments, is I actually sit down and I'll go through either Colorado's guidance and like, try to create one from scratch for myself and answer it on partly on like a fictitious client in the ad tech space. Because I find that by going through that exercise myself, it kind of opens your mind up to, okay, this is what we really are getting at. Because usually in the ad space, to your earlier point, there's really only a relatively small number of things that can really go sideways on you, but you need to make sure that you really cover off on those things. Because if a regulator were ever to look and you hadn't covered off on those things, you're in big, big trouble.
- [22:06]
Omer Tenney
And I think it's not just a regulator. This kind of goes back to a more basic privacy conversation. To me, the most important regulator is the market and consumer expectations. And if you are wielding technological tools in a way that's creepy and that kind of turns consumers against you, it's not a good business decision. Right. And you'll get bad coverage in the press and people won't like it and will kind of push back against it. So I think when I talk to clients, and a lot of our clients are technology startups actually, so, so I talk not to, you know, someone in the compliance kind of division. I talk to the, the founder, the, the CEO, and I tell them, you know, think about the market reaction. Are you doing something that will seem kind of untoward, sleazy, creepy, or will it land kind of, well with the market. And a lot of it, I think, is about transparency and explaining stuff. So, so oftentimes I focus not on the privacy policy, but on the FAQs, like tell people what it is you're doing. And apparently you're doing it because you think there's a value prop there. So convince the public and then you'll be fine.
- [23:31]
Alan Chappelle
Yeah, that's always an interesting conversation, particularly when you're dealing with a founder and particularly when it's a new relationship. Because sometimes you have to tell the founder that his baby is ugly. And very few people want to hear that.
- [23:46]
Omer Tenney
That's not our gig, Alan. Maybe you didn't do that, but we, we try to be a bit more diplomatic and big law, and the baby is usually you can find nice things about it.
- [23:58]
Alan Chappelle
Fair enough. I, yeah, I tend to get away with things that big law does not. On the other hand, you guys have associates. And boy, there are days where I would give my right arm for a couple of, couple of really, really good associates.
- [24:11]
Omer Tenney
That's the secret right there. That's the secret sauce.
- [24:15]
Alan Chappelle
Absolutely. So I got one more question for you. So, you know, lawyers trying to understand AI, you know, use within, you know, their clients, companies, and they often struggle to communicate with their tech and product counterparts. And, and, and so, okay, we can take your baby is ugly off the table. Like, you know, what's your advice for bridging that gap?
- [24:37]
Omer Tenney
It's a very different conversation, right. When we talk with lawyers and certainly privacy lawyers, for us, that it's even different than talking to the gc. Right. There's like a million different issues, and privacy is one of them. Privacy, probably a small one of them, but certainly talking with technology people or business people. And my advice is leave the law out of it. It's not about the law. Forget like, you know, section this and that of the ccpa. Privacy is about public perceptions. That's what it's about. It's not a compliance issue. It's a market reception, consumer expectation issue. And that's the mindset I think you should be in. A lot of technology people, you know, if you kind of live in Silicon Valley and kind of breathe that entrepreneurship kind of ideas, business models that could land, you know, not so well with the general public. So how will your grandmother kind of react to this new technology? How will parents of kids who go with your kids to school think about it that way? And I think if you implement that, you'll be okay.
- [25:56]
Alan Chappelle
So my last question is, and we do this not with every guest, but with many guests, we want to know a little bit more about the person behind the great big brain. And so what's your semi secret hobby or passion for lawyers?
- [26:12]
Omer Tenney
You know, in big law, there are no hobbies and passions because we have to bill every minute of the day of the week. But having said that, I do like skiing. Unfortunately, this has been a very real winter up in New England. So I'm gonna go tomorrow. We're recording this on Friday. I'm gonna go up to the mountains tomorrow. We haven't even decided where, but we can go to Main or Ver on. So big skiing family.
- [26:45]
Alan Chappelle
Yeah, that, that's a good one. And I, I haven't skied as much. We lived in Utah back when I was in, in late high school and college. Well, yeah, so we were right by Snowbird and Alta. I kind of got spoiled and suddenly those little hills on, on the east coast seemed less cool. I have similar issues with friends who were, you know, born in Europe and have like gone and done, you know, gone and skied in Switzerland. And they just tend to look at everything we have here as like, oh, that's very, very nice.
- [27:16]
Omer Tenney
Alta and Snowbird do give Switzerland a run for it, so I wouldn't knock them. But yeah, compared to east coast skiing, obviously it's not even skiing, but you know, we can drive up, it's nice.
- [27:29]
Alan Chappelle
But the coolest thing about skiing, and this may get back to the whole lawyer business, is something like that requires a level of focus but also relaxation in order to do it. And anything that you can do, do that opens your mind up and gets you thinking about not what you're dealing with on a day to day basis is the biggest gift in the world.
- [27:48]
Omer Tenney
Love that as, as an ending, Alan, especially coming from a musician like you. So thank you. Yeah, I agree wholeheartedly.
- [27:58]
Alan Chappelle
Omer, thank you for so much for coming on the pod. This was wonderful.
- [28:02]
Omer Tenney
Yeah, thanks for having me, Alan.
- [28:06]
Alan Chappelle
That was a really fun conversation. We've got a bunch of other fantastic guests coming up on the Monopoly Report podcast over the next few weeks. Please subscribe to the show@monopolyreportpod.com or on Spotify, Apple, YouTube, or wherever you listen to your podcasts. Thanks for listening.
- [28:27]
Omer Tenney
Thank you for listening to the marketecture podcast. New episodes come out every Friday and an insightful vendor interview is published each Monday. You can subscribe to our library of hundreds of executive interviews at marketecture tv. You can also sign up for free for our weekly newsletter with my original strategic insights on the week's News at News Market tv. And if you're feeling social, we operate a vibrant slack community that you can apply to join@adtech God.com.