[00:00] Debbie Reynolds: The personal views expressed by our podcast guests are their own and are not legal advice or official statements by their organizations.
[00:11] Hello, my name is Debbie Reynolds. They call me the Data Diva. This is the Data Diva Talks Privacy podcast, where we discuss data privacy issues with industry leaders around the world with information that businesses need to know.
[00:24] Now I have a very, very special guest, a dear friend, longtime colleague, collaborator, in good trouble. We like good trouble, right?
[00:33] Heidi Saas. She is a data privacy and technology attorney. Welcome.
[00:39] Heidi Saas: Thank you so much for having me. It's been quite a while since this interview session. I believe I was episode number 76.
[00:47] Debbie Reynolds: Well, we're in the three hundreds now. It's been a minute. Yeah. So we have a lot, a lot to talk about. And so I highly recommend people go back and listen to Heidi's initial.
[00:58] So this would be kind of part two. So we're catching up, figuring out, like, what her world is like.
[01:04] I feel like the world's changed a lot in the last few years.
[01:08] I still feel like sometimes we're screaming into the void around privacy, but I think people are listening more. So I feel like the media is picking up these stories more, especially as these states are passing more laws.
[01:21] We're literally seeing some of the harms that we've been preaching about for years literally come to pass. So I think that creates more conversation. But tell me, what is going on in your world?
[01:34] Heidi Saas: Well, thanks for having me. These are my personal views, not legal advice.
[01:38] It's always a pleasure to have a chat with you and check in and see what your perspective is, because we're seeing the same things, and you and I often see them the same way.
[01:48] And I'm surprised the impatience hadn't gotten to us, because you're right, we've been talking about the same things for quite a while, and I think we have made progress in a couple of areas.
[01:57] I think the first is awareness.
[01:59] More people are aware of the power of data that is advantageous power and discriminatory,
[02:12] neglectful,
[02:14] all sorts of dangerous other uses of data and the power associated with that. So it depends on where people are in terms of feeling powerless.
[02:26] I don't think that is the majority opinion of consumers now, because they do know that they have settings, and those settings have impacts with the companies,
[02:38] and they do have the option not to use some of these tools, but they don't want to be left out of their communities.
[02:46] So they're starting new communities that don't have technology at the forefront. And I think that is another new thing, too, is the new Hotness of in person activities for humans,
[03:01] that's like a thing. I think the third thing I'm going to go with is technology is getting us closer to where we want to be, whether they wanted to build for this or not.
[03:11] Because the original purposes of the, you know, ad tech, Rails,
[03:16] supporting the open web, it's based on surveillance. People don't want to be tracked. And so what else are people doing? There are walled gardens. Well, we don't necessarily like that either.
[03:26] So people are pushing, you know, the technological boundaries to find new things that are closer to the promises that were initially made when these technologies were introduced.
[03:37] So I think one of those examples is Apple is starting to use edge computing to do facial recognition for age gating, I believe.
[03:47] And they stopped allowing cross device tracking and identity resolution companies to follow people on their iOS supported devices in Safari.
[04:02] So if you remember the ATT, that had a pretty big impact and now this. So it looks like Apple's our most effective regulator in privacy globally these days because they're making business decisions and we see you and we know you want privacy and we kind of need to do something about it every now and then or we can't use this as our marketing.
[04:21] So props to them for that. But I think those are kind of the few areas where we do have a difference now as opposed to then. So like the awareness and people's empowerment and the technological advances that we have as opposed to where we were previously.
[04:37] Right.
[04:38] Debbie Reynolds: I think one of the surprising things that has happened and you touched on a little bit is that we're seeing in my view more how a consumer's voice and how their voice, especially if they together say, hey, I don't like this.
[04:56] They are changing behavior because companies want to sell products. Right. They want to make money. And so we're seeing even outside of regulation. So two things I think that are happening in the background, I didn't expect as much.
[05:10] So one is certain companies are putting pressure on other companies to try to fall in line with privacy. And so Apple's a good example of this with like things like app transparency,
[05:23] them being an OS maker, they can push down a lot to their ecosystem and third parties. But then also we're seeing and talk just touch a little bit on the GM car article that Kashmir Hill did in the New York Times.
[05:40] And so that had a huge impact because consumers were like literally I was doing at that time, I was doing my report for the US Department of Commerce on IoT and I read her article, I read all the comments and Just the things that people would say, like, I'm a third generation GM car buyer and if they're selling my data,
[06:04] I'm never going to use them again. And it's like,
[06:07] why would you lose a dollar to make a dime,
[06:12] literally.
[06:13] So I think that that message is coming across, you know, maybe not as much as I would like, but I think that that is turning the tide a bit because I am hearing more stories about customer backlash and companies really responding to that.
[06:31] Heidi Saas: I hear you and I agree with that. So GM was in the news today. Did you see
[06:37] Debbie Reynolds: what they do now?
[06:39] Heidi Saas: So a bunch of the states went after them as well and they reached, I think $220 million settlements.
[06:48] It was announced yesterday, I believe, I believe Oregon was the state leading that settlement.
[06:55] So I know you know this fun little story, but I'm going to share it with the public here.
[07:00] I met this nice man after a public speaking event in 2024 and he said, I can't get insurance and my car company won't tell me why and they're sharing an insurance company and they won't tell me where I gave them consent to do this and I've got a whole group of people and they got the same problem.
[07:19] And I said, wow,
[07:21] number one, I'm not licensed in your jurisdiction. Number two,
[07:24] it sounds like you need to sue everybody to find out what happened. Let's get you to qualified counsel in your area so you can talk to them about what's going on.
[07:33] And number three, please report this to your consumer protection division.
[07:39] And so very nice man, stayed in touch and is still in my network and small business owner and I've seen him in what he is doing with his business includes more and more of data.
[07:51] And so this was a learning experience on a personal and professional level for this individual.
[07:57] But after we connected six weeks later,
[07:59] that is when Kashmir Hill reached out and spoke with them and that was the first GM class action lawsuit filed.
[08:07] So I lit that fire. It took 10 minutes of my time.
[08:11] It was not legal advice, it was just consumer protection issues that were being highlighted,
[08:17] like people tell me all the time.
[08:19] So I'm excited to see how big that fire got because the FTC took a piece and. And two weeks after this lawsuit was announced in the New York Times, the GM said, gee, that's my bad.
[08:32] I didn't know about that. We're not going to share that data anymore. And Lexis is like, yeah, we're not going to do that either.
[08:37] And then the other states started picking it up and it was all of the players in the supply chain of that ill gotten data started to get regulatory heat and it lasted.
[08:47] That was 2024. It's the end of 2026 and they're still settling with the states.
[08:53] Untold billions of dollars it has cost them for doing privacy the wrong way. They tried to say they got consent three different ways. None of it was effective. As if the consumer is that confused, wondering, how did I give you consent to take over all of my data and share it with all these people so I would be denied services later in life.
[09:13] I didn't consent to that.
[09:15] So I mean, there are a lot of lessons to be learned here.
[09:18] Most of all, like you should listen to what consumers are complaining about instead of trying to optimize everything.
[09:27] Just like you said.
[09:28] Why would you try all of this just to make an extra dollar?
[09:32] You have record revenues going right now if you're not burning up all your cash with the AI hype cycle.
[09:38] So like, why don't you spend more time listening to the humans that you're trying to do business with?
[09:43] Maybe you'll have, you'll mitigate some risk that way in terms of litigation, but, but also you'll build more trust. And we have a crisis of trust in our country. In every single area that you look right now,
[09:54] disinformation and distrust is being created left and right so that people can move into a vacuum to assume power.
[10:01] I don't want that to be our future.
[10:04] Debbie Reynolds: I want your thoughts on the Fair Credit Reporting act. And I think you may have posted about this and I always thought maybe this could work.
[10:15] I want you to either talk me into it or out of it.
[10:18] And that is,
[10:19] I have thought and I hadn't thought it all the way through,
[10:24] but I thought that data brokers should be regulated under the Fair Credit Reporting act because decisions are being made about people based on the things that they're doing and saying.
[10:37] And then that information, just like you said, this gentleman who approached you about cars is not transparent to the person.
[10:45] Right. And so that's what the Fair Credit Reporting act was supposed to do, was create transparency. Right. And so now we have this other industry underneath this gargantuan that is making these decisions about people and it's not, not transparent.
[11:03] But what do you think?
[11:04] Heidi Saas: 100% I agree. 100%.
[11:08] This sounds so dorky, but I gotta do it.
[11:11] I did a TED talk and in my TED talk I covered this very topic.
[11:16] But I've previously done public speaking events for years trying to get people to Understand why I hate data brokers?
[11:25] It's because our lack of agency and power over our own data has put us into these positions where we no longer have leverage.
[11:32] The Fair Credit Reporting act was enacted in 1970,
[11:37] and it defined the credit reporting agencies in a narrow type of way, saying that they created credit reports about people at that time as the only place you could go to get that kind of information about people.
[11:51] So Congress saw what was happening and said, ooh, this could be a problem.
[11:56] So they passed the Fair Credit Reporting act and it includes a private right of action.
[12:01] And that tells you how important it is for, for them to get this right.
[12:05] Now it only covers a few areas.
[12:08] So housing, employment,
[12:10] lending,
[12:11] education and insurance, those are the only areas. But if you think about it, those are fundamental areas of our lives,
[12:19] right?
[12:20] So it matters what data they're looking at when they're trying to determine if you are qualified for the job that you applied for.
[12:30] What doesn't matter is the shopping cart contents on your website, in your cookies.
[12:36] What doesn't matter is all of the people that you have in your social graph as potential relatives and maybe what their political leanings are like. None of these things are relevant to your skills and whether you're qualified to do this job.
[12:51] The Fair Credit Reporting act is being used. Mobley vs Workday is a case that, that is currently a huge class action.
[13:00] And they're gathering data and information from the class members at this point.
[13:07] Excuse me.
[13:09] And some of that information is what jobs did you apply to so that they could go back and follow through the technical and digital forensics to see what happened?
[13:21] Because what's alleged is that applicants over 40 and applicants of color were not referred for jobs. They just, their resumes just went into the trash.
[13:33] And so that's a problem. Right?
[13:36] And so they wanted to, you know, Workday wanted to kick it around and say, it's not our fault, it's their fault, and it's this data. Now we have to get into what data did they use,
[13:43] what scores were assigned to the different data pieces,
[13:48] and how were they categorized to find out was discrimination hooked into the system. It doesn't have to be intentional. It's disparate impact.
[13:57] So it's not like you've got racist coders sitting and developing tools that are going to squeeze people out of jobs, right?
[14:05] That it's just the way the tools work and they're using too much data.
[14:11] So under the Fair Credit Reporting act, you probably have access to a couple of data,
[14:15] couple of data points, things are relevant with the information from data aggregators and data brokers and the hiring systems and ATS tracking systems that are being used right now.
[14:26] It is an ocean of data.
[14:27] Everything on the Internet must be true, right?
[14:30] So, yeah, people aren't getting jobs or loans or apartments or insurance.
[14:36] So the Fair Credit Reporting act is being used. The courts are very slow, and companies do not want to give up the information or let people into their tools. Not researchers, not plaintiffs counsel in discovery.
[14:47] They don't want anybody into their tools. The opacity is their cover.
[14:51] Debbie Reynolds: Right.
[14:52] Well, two things that give me some hope here that I'm glad to see is that I'm seeing certain jurisdictions band the definition of what they consider to be a data broker.
[15:05] Because some companies say, hey, I'm not a data broker, but you're actually doing things that data brokers do. So defining that. And then also,
[15:13] a lot of the things that we're talking about are captured in digital Systems where maybe 50 years ago they weren't right. So it may have been harder to show that there was, you know, desperate impact.
[15:26] There was, you know, discrimination. There was all this other type of stuff, but it's in the code now. It's in the data now. And so that data is there and is there for people to look at, make sure things are done right or not.
[15:41] What do you think
[15:42] Heidi Saas: independent audits are a thing? Like they should happen? I mean, they're not going to happen until they're mandated. And so somebody's got to go first and say, listen, all of these tools were built.
[15:51] None of them were tested. They certainly weren't tested for fairness.
[15:54] So we need to get through them and check this, because technology is not going to get any easier. There's no finish line. So if we don't get this right now,
[16:03] as we're moving into the threshold where everything is done online, if we don't get this right now, we're never going to be able to get this right.
[16:11] So I've been working with 4Humanity for over five years drafting the criteria for how to conduct independent audits of these tools. We started with employment tools. We based it on New York's Rule 144,
[16:26] until the teeth were lobbied right out of it by the industry,
[16:29] as happens.
[16:31] And then we shifted and focused more on the EU's AI act, which is now being lobbied and attacked in Europe.
[16:39] So all of that messy business aside, the criteria are the same. It's pass fail, it's objective.
[16:45] It's one of those situations where you need to ask questions what are the different use cases in this technology?
[16:51] Right. Get to take apart the tool. Where is the data coming from? What is your tool doing with the data? Who are the stakeholders?
[16:59] What are the possible rights that may be impacted? What are the risks,
[17:04] treatments and mitigations for such?
[17:07] These are critical thinking exercises that people need to do as part of their AI governance programs.
[17:15] If you do privacy, you've probably been voluntold to include AI governance as well.
[17:20] But these are the same sorts of considerations that you would put into your governance program.
[17:25] But it's tool by tool, especially when there's a high risk of potential civil rights harm.
[17:33] That's the employment tools, all of them.
[17:36] It's not every single tool needs to have these sorts of independent audits,
[17:40] but these in these life critical areas most certainly need to have these audits done. And they need to go on a regular basis. They can just tack them on along with the cybersecurity audits.
[17:51] Because if you're already in there and you're testing different functionalities for your systems,
[17:56] why not include both?
[17:59] Right.
[18:00] Debbie Reynolds: What do you feel makes AI different than some of these other technologies? And I can give you my thoughts.
[18:09] Heidi Saas: Well, are we talking about generative AI versus DE learning and other
[18:14] Debbie Reynolds: just AI in general?
[18:16] I think the way that people are using it and the way the technologies are going.
[18:20] So I feel like the issue with AI,
[18:27] there are several issues. Not that it's a horrible technology or anything, but the thing that concerned me about this and other emerging technologies, they are very similar where the harm can be very rapid and it's something for which there may not be adequate legal redress.
[18:47] Right.
[18:48] So let's say you went for operation and some AI bot decided that you needed your toe activated and then they said, oh, we made a mistake. It's like, well, you can't undo that.
[19:02] So that is the thing that concerns me a lot. When companies are leaning too heavily on that without having that human,
[19:10] I try to say human in the lead as opposed to human in the loop. Because when you're in the loop, it doesn't say what your role is in general, but what do you think
[19:20] Heidi Saas: accountability is where I'm at,
[19:23] I think personal accountability would do a lot because if a individual had done some of these things that these companies have done, they would be in jail.
[19:33] But because they were companies,
[19:35] they are not held to the same behavioral rules as individuals are.
[19:42] There are criminal companies.
[19:44] Right.
[19:46] So I think accountability is important.
[19:49] If we had more executives charged with crimes,
[19:55] whether they were convicted and went all the way to Jail or not,
[20:00] just charged with crimes in plausible cases where the evidence is clear and call it a crime instead of corporate wrongdoing.
[20:11] It's crime. It's theft, it's fraud, it's crime.
[20:15] I think that would set a big example.
[20:17] Like, not everybody's a Bernie Madoff here,
[20:21] but the testing and production is risky for businesses now because it should be.
[20:28] I know it saves you money and it saves you time on product market fit,
[20:33] but your agile development system is dangerous in some circles.
[20:39] The harms are beyond what we can pay for at this point. If we look at Meta,
[20:47] how many parents do you see standing on the steps of the courthouses holding pictures of children that are no longer with us? Those children died,
[20:56] most of them in the homes with while the parents were there.
[21:00] And they had no idea that their life was being destroyed and they would take their own lives in the next room.
[21:07] How many parents did you see on the steps of the courthouse? Right.
[21:11] I mean, he even stood up when he went to Capitol Hill and apologized to the parents in the back of the room. He acknowledged the harm that he had been causing by putting profits ahead of safety.
[21:21] But did he stop?
[21:23] No,
[21:24] he did not.
[21:25] And so, and not only, you know, did they have juries say, you did this.
[21:31] The judges themselves said, your executives are liars. Look at the public statements they made in the emails they sent to each other. You're liars. That's a huge,
[21:40] huge shaming attempt by the court to say this behavior is not okay. And what happened? Their stock price went up.
[21:49] Yeah.
[21:50] So, I mean, we need more accountability for bad behavior,
[21:54] especially if we have to go to this far of a length and effort to prove the bad behavior. Despite tens of thousands of lawyers and hundreds of millions of dollars trying to crush anyone calling into question their business decisions or bringing any of this evidence to light,
[22:13] we still found out the truth as truth came out.
[22:16] And so what change is being made?
[22:18] Nothing.
[22:19] Now they're coming out with cute little toys and saying they're the leader in children's safety now because of the things they agreed to in a settlement.
[22:28] So I think we need more accountability before the bad behavior stops. Shame isn't enough.
[22:33] Not even sure, you know, like crippling financial loss at this point. It's not even at a scale enough to cause him those problems. It just keeps going.
[22:43] The lawyers got in trouble because they were declaring everything was attorney client privileged, and the court went after the lawyers for being shady in this.
[22:51] It's shameless all the way down.
[22:53] Debbie Reynolds: Shameless all the way down. I think that's true.
[22:56] We're coming up on Christmas here.
[22:58] What types of things do you think?
[23:02] Just everybody. Not just, you know, kids, parents, aunts, uncles, people going out buying things.
[23:10] They're. I'm sure they're getting all these marketing prompts to be able to buy stuff or their friends are trying to get them to buy different things. What things do they need to be thinking about?
[23:24] Heidi Saas: So Liveramp is now on GPT.
[23:27] So. Yes.
[23:29] So you are getting personalized targeted advertising Even in your LLMs using your prompts,
[23:38] you're getting targeted advertising everywhere.
[23:40] So think about that first of all, like, what are you buying? Who are you buying it for? Why are you buying it for them? Do they even want that?
[23:47] Or was that just some idea?
[23:50] Now next,
[23:51] what is it you're going to buy for them? Is it a digitally connected device? Because sometimes you can't even get a yo yo without having an app attached. Like what is that about?
[24:00] Do you really want to burden someone with, with a gift that you got for them that also includes a privacy invasive app on their phone? Like, think about more than just the gift itself.
[24:13] Cash is great,
[24:15] right? I think cash is underrated. People always love it. It's the right color, it's the right size, nobody ever gives it back. I think that's an option as well. But I would caution people against the digital tools.
[24:27] And I think the scariest kind of digital tools I've seen lately are the intimate digital tools.
[24:32] So please,
[24:33] please, I know a long distance relationship may be hard,
[24:36] but talk about irreparable harm, man.
[24:39] Like I can't this. Don't just know people don't do that. Get a plane ticket.
[24:46] Okay? So that's what you could do.
[24:48] The beauty of in person events now, something I like to do with my family just because we'd like to focus on things that are not money based.
[24:59] And this may sound a little cheesy for people, but that's fine. You can judge me if you want. I'm happy where I'm at.
[25:04] But we like to do events,
[25:06] so we will give gifts to each other that are kind of an experience. Like something might be a kayaking trip on this river in this month of the year and then we'll go and put it on the calendar or something like that.
[25:20] Or it could be I got this cookbook at the Goodwill and It was only $3, but I picked out these four recipes that I wanna spend the time cooking with you for while we're home for winter break or something like that.
[25:32] So it involves an in person experience. And so some of the things that I gave to them were,
[25:38] I will set up zoom calls for you three times during the Christmas break so you can still check in with your friends on a video call, and all of you at the same time can be there.
[25:49] Another thing, last year or the year before, we got them a landline,
[25:53] a phone, an actual phone so that they could call and talk to their friends.
[25:57] And their friends were immediately like, well, what kind of games do you have on your phone? So then we had to explain what landlines were, and it started to spread.
[26:03] The other parents were like, that's a great idea. And so they started to do the same thing and get those smartphones out of the hands of their children.
[26:10] And the last thing I did was I said, you know what? Over the summer break, my gift to you is that I will host your D and D club over the summer because you're not at school and I know you miss having your club, so.
[26:22] So send out the message to your club members and I'll put it on my schedule now that once a month, I'll do your club meetings and I'll supervise them here at the house.
[26:30] And we did so.
[26:33] Which made for some interesting client calls. But they were ever so gracious. And I am grateful for my clients understanding that while I'm giving advice on a call, I also have, like, half dozen teenagers in the other room doing a D and D campaign.
[26:45] So. But the in person event, I think was one of the best gifts that I got for the kids because they got to be with their friends in person, and there wasn't a cell ph phone to be seen.
[26:56] Debbie Reynolds: Oh, my goodness. I think that's a great idea. The. Who me?
[27:01] Heidi Saas: Yeah. What do you think about what people should be getting for Christmas and what we should stay away from?
[27:08] Debbie Reynolds: Maybe it's because of my ilt work that I've done. So I look at what I think is, like, horribly harmful,
[27:15] and I decide, okay, I'm gonna spend my time, work on this for a bit and do what I can to help that area and then move on to the next thing.
[27:24] Pretty much anything that's connected to the Internet concerns me.
[27:28] So I'm like, why is it connected to the Internet? You and I, we were in Washington, D.C. when I visited you there a couple years ago that. That you were renting the privacy house.
[27:39] We saw that the range hood over the stove, and they had, like, the little symbol for the Internet, the Bluetooth. We were, like,
[27:47] Heidi Saas: sure that it had a Bluetooth on the range Hood. We were
[27:49] Debbie Reynolds: like, we're like, what?
[27:51] Heidi Saas: I know, why people why,
[27:55] Debbie Reynolds: right? And like I was shopping for a washer and dryer not long ago and they're all connected. Obviously you can choose not to have them connect to the Internet, but I feel like anything that's connected to the Internet you need to be leery of.
[28:09] Like
[28:09] Heidi Saas: a freezer. I believe the French had that cyber security attack with the grocery stores where they hacked in and they cranked up all their freezers and ruined all their food. They didn't go for the money, they went for the controls of the freezer because your freezer shouldn't be online.
[28:23] Debbie Reynolds: But
[28:23] Heidi Saas: yeah, well, when it comes to Christmas, like, I can't imagine being you. All the work you've done on Iot and the guidelines and frameworks and all that stuff, the advice you've given, like, I can't imagine like you buying anything Iot for anybody.
[28:36] Debbie Reynolds: No, no, no. Everything's unplugged. Nothing is going back old fashioned. One thing I will say,
[28:43] I think people should do, and this, maybe this sounds cheesy, spend some time with your family or your friends,
[28:51] maybe one on one and try not to look at your phone.
[28:55] Heidi Saas: I love that. Right? I love that,
[28:58] Debbie Reynolds: like some undivided attention to your family members. I think they were really, really enjoy it and you may as well.
[29:05] I remember one time I was in Switzerland, I was in Lucerne, Switzerland, like overlooking the out Alps and the river.
[29:14] And I was just looking around in the park because it was overlooking the water and everything. And then I'm sure I was the only American there at that point, but they were,
[29:25] people were holding hands, they were talking, they were like playing ball and I was like the only idiot with a phone.
[29:32] And I thought, what is my problem?
[29:35] Like this is absolutely gorgeous. It's like this is what life is supposed to be. Supposed to be you enjoying your time, not trying to see what's the next alert or what's the next notification.
[29:47] I don't need to know who won the Kentucky Derby. That's not important. That's not going to be important a hundred years from now. So I try to focus on what's going to be important in the future,
[29:58] not the thing that's like buzzing and trying to get my attention at the moment.
[30:02] Heidi Saas: Yeah, I think so officially, our advice for the holiday season, the two of us, is let your presence be the present.
[30:10] Debbie Reynolds: That's right, that's right. I agree with that. I agree with that.
[30:14] Well, what is it on the privacy landscape that we need to be Thinking about going forward right now,
[30:24] Heidi Saas: privacy litigation is not going to slow down.
[30:27] So there were legislative changes in California that removed one of the SIBO causes of action. There are still two others.
[30:34] And also **** wasn't even the most alleged cause of action over the time period that was studied in a recent report that we saw about litigation trends.
[30:45] So there are several causes of action that are still going strong.
[30:49] And I think it's premature for people to think that one legislative fix is enough and maybe we can get another legislative fix later.
[30:59] That's a whack a mole mindset. I think the better idea is for you to, number one,
[31:04] take your customer journey through your consent tracking process. If you have a cookie banner, get in there and click the buttons. If you have a do not sell,
[31:13] do that. But you need to test what your program is doing and you need to have the technical wherewithal yourself or access to the engineers that can help you look at your data traffic and tell you when your pixels are firing,
[31:28] where your data is going to. The answers to these questions are necessary for you to be able to do your job and mitigate risk.
[31:36] Also to make sure that your proxy program is being honored. Right, because you can say one thing, but your pixels are firing and you know it is different from what you have disclosed to the public.
[31:48] That's unfair and deceptive.
[31:50] So it's a problem. So take that journey yourself. If you're working in the business, don't just set it up and trust that you know it says, yeah, yeah, it's working.
[31:56] Like take the journey yourself and see.
[31:59] But also, I think that you should think more about the data that you're using in terms of data minimization. I'm going to ask you one more time. Rethink what data you're collecting.
[32:11] It can't be breached if you never collected it. There are a lot of different technologies that are out there that you could deploy that you don't need to collect so much personal information.
[32:21] Look at the nature of your use of that data.
[32:24] If you don't need it, don't collect it.
[32:27] And honestly, purpose limitations are contractual and contracts don't stop data flows. So if you're working in technology and you have some ideas of technologies that work in this field,
[32:42] I want technical controls for purpose limitations. And if you're working on that, please let me know. Send me your research papers. Reach out to me, find me on LinkedIn, let me know.
[32:50] This has been bothering me for a couple of years and I keep talking to engineers about it and nobody wants to do this,
[32:57] but everybody recognizes that this would solve a lot of our problems if there were a technical way for data to not go where it's not supposed to be. So it doesn't have an ability to leak.
[33:09] It encrypts itself at the point where it recognizes it's at a illegitimate ISP or something. There's gotta be away. I don't just want things like, there's a reason why I want these things.
[33:20] And I'm sure that I'm going to find the right engineers one day.
[33:25] Right.
[33:25] So I might as well ask for them now.
[33:28] Debbie Reynolds: Right. Because data takes a journey. And data flows and paper and laws and regulation don't stop flow. Data flows, yes.
[33:36] Heidi Saas: Machine unlearning is not a thing. And LLM memory. Yeah.
[33:40] Yes, it's not a thing. You could come at me if you want, people, but it's not a thing. And LLM memory is a real problem. Businesses are losing trade secret intellectual property and process flows, workflows,
[33:53] how their employees are using critical thinking skills,
[33:56] game plans, strategies. All these things are going into the LLMs. People are just puking them in there and they're not thinking about that as a loss.
[34:04] You're just giving it up, you know, and you can try as hard as you want to de identify data, but the LLMs are very good at using artifacts and architectural infrastructures to recreate what that identity was and re identify.
[34:17] Even though you took the effort to strip identity. It's coming back. LLM memory is going to be the next technological challenge if we want privacy to succeed, especially in the frontier models.
[34:27] But you and I both know the future is with the small models, right?
[34:30] Debbie Reynolds: Oh, absolutely. I've been saying that forever.
[34:33] It's all in the small models. Right. And really that's what.
[34:38] Before people went gaga over generative AI LLMs, that's really what AI was. So AI was very purpose built,
[34:47] very specific and narrow so they couldn't get into trouble because
[34:51] Heidi Saas: it was pretty good.
[34:52] Debbie Reynolds: It was very good. It was very good at what it did.
[34:55] Heidi Saas: Yeah. Humans still have the ability to make decisions like, yeah, I like that. Let's get a little bit back to that. Because right now everybody's kind of building the buffer layer in between whatever you're doing and what the frontier models are doing.
[35:07] And like,
[35:08] I don't know that that's where we need to focus our time. It's kind of like saying the parents aren't doing a good enough job of watching their kids while the people who are creating the tools or know they're making addictive tools.
[35:18] Debbie Reynolds: Yeah, that's true. That's true.
[35:20] Heidi Saas: It's Gaslighting is what that is. So
[35:22] Debbie Reynolds: it is
[35:24] Heidi Saas: on. Gaslighting is what we're doing, y'. All. Thank you so much for that.
[35:29] Debbie Reynolds: Well, I'm so happy to be able to chat with you and I'm so proud of all the work that you do and you're just an important lady. You're an important lady and you're doing some important things.
[35:39] It'll be things that you're doing now are going to be important 100 years from now. So that's really the best that can ask for.
[35:47] Heidi Saas: And you, too. I wouldn't be doing this if it weren't for you. Right. Thank you so much. All right, well, that's
[35:54] Debbie Reynolds: a wrap. Thank you so much for being on the show and it's been great chatting with people. People definitely follow Heidi. She always has some great hot takes that she does and she drops a lot of knowledge on LinkedIn.
[36:06] Heidi Saas: Thank you so much, Debbie.
[36:09] Debbie Reynolds: All right, talk to you soon.