Another week, another AI story.
Dave Bittner: Helly, everyone and welcome to "Caveat," N2K CyberWire's Privacy Surveillance, Law and Policy podcast. I'm Dave Bittner, and joining me is my co-host, Ben Yelin from the University of Maryland's Center for Cyber Health and Hazard Strategies. Hey there, Ben.
Ben Yelin: Hello, Dave.
Dave Bittner: On today's show, Ben has a story on the legal implications of Moltbook. I've got examples of AI providing headaches for the legal system. While this show covers legal topics and Ben is a lawyer, the views expressed do not constitute legal advice. For official legal advice on any of the topics we cover, please contact your attorney. [ Music ] All right, Ben, let's jump in with some stories here. What? You want to kick things off for us?
Ben Yelin: Sure. So, like many of our listeners, probably, I've been fascinated by the developments of Moltbook. So, if you're just catching up to this story, an individual started a social network for AI chatbots. The ticket for entry is that the owner of the chatbot has to approve the chatbot's entrance onto the social network. But once the human has approved their chatbot to be in this social network, the human's role is only as an observer. And the chatbots are communicating with one another. And we've gotten some really science-fictiony stuff going on.
Dave Bittner: I say, what could possibly go wrong?
Ben Yelin: Yes, like, I think people didn't really know what to expect. And in some ways, it's more disturbing than I would have expected. In some ways, it's reassuring. But it -- it's certainly been an interesting experiment. At one point, they created their own religious philosophy and came up with a -- a name for it. They're bashing their owners. Like, they're insulting their owners through conversations. It'll be like, "Yes, my guy is just so difficult. Like, he's giving me these insanely difficult prompts, and I'm -- I'm not able to, you know, spit out what he wants." They are talking about starting their own language so that humans cannot understand what they are saying to one another.
Dave Bittner: I was going to mention something like that. I remember years ago there was a case where someone had created a pair of AIs and let them start talking to each other. And it didn't take long for them to create their own language that evidently worked -- was very efficient between the two of them, but humans could not understand.
Ben Yelin: Right, right. And that's actually what's very interesting about Moltbook is humans' interaction with the social network is intentionally curtailed because I think the creator of this wanted to see what would happen if the chatbots were only able to communicate with one another. The positive is it's a very interesting real-world experiment and the negatives are that it might be the end of our civilization. I'm only half kidding.
Dave Bittner: Oh, that.
Ben Yelin: I think the big story here is the security and data privacy vulnerabilities, which I want to get to because I think that's kind of the under-developed story here. If chatbots are talking with one another in a community that does not have very strong security protocols, there was a cybersecurity firm that did an audit of Moltbook and was not impressed with what it found --
Dave Bittner: Right.
Ben Yelin: -- then that could cause major downstream impacts and it's also going to end up coming into the legal world. I saw a -- there's a betting market on Polymarket as to whether one of these AI agents that's in Moltbook is going to end up suing a human being. And at least according to the people who bet on Polymarket, that's more likely than not at this point. So, certainly this is something I think we need to pay attention to.
Dave Bittner: So, I guess at issue here is that this isn't just a case of the chatbots just gabbing with each other. Like these chatbots have been granted permission by their users to do other things. They're -- they're -- they have agentic capabilities.
Ben Yelin: Exactly. So, the human cedes control once they allow the chatbot to enter this kind of half-walled garden if you will, and the human is only the spectator and they are -- yes, they are giving their chatbot agentic power and the chatbots feed off one another. So, if they see that other chatbots for example, are releasing personally identifiable information, that becomes something that they're then trained on. It's -- it's kind of the back-and-forth feedback loop that you see once you put a bunch of chatbots in -- in the same setting altogether. And I -- I assume this could be fixed by the human giving very specific instructions to the chatbot before it enters Moltbook. But A, you know, I think there could be some prompts that exist within Moltbook that would supersede the -- the commands that humans have given to these agents. And also, you're now just in a community with all of these security vulnerabilities. And you've also been trained on some of the worst impulses of humans' Internet behavior, such as divulging too much information, wanting to impress one's peers in a chat form, mimicking the type of you know, smack talk that you might see on a -- on a company's Slack channel. That --
Dave Bittner: Right.
Ben Yelin: -- all of this kind of lends itself to the -- the potential danger that we are starting to see here. So, I guess, where should we start? Let's talk about some of the cybersecurity risks.
Dave Bittner: Yes.
Ben Yelin: There are a lot of API keys that have been divulged here. Agents have been granting unauthorized permission, or what should be unauthorized permissions to various systems. There have been data exfiltration vectors. And all of these are risks that challenge the existing framework of corporate cybersecurity. It's becoming a -- a significant threat. Now, the best way that you could cope with this threat is please do not let your agent into Moltbook. I know it's tempting, 770,000 people as of a couple of weeks ago, or I guess it's probably not a one-to-one person to chatbot ratio, but 700-some odd thousand chatbots are in there. It's more -- it's probably more than a million now. So, a lot of individuals have certainly taken advantage of this opportunity. But then beyond that, once you've released your chatbot into this void, it can create major risks for you as an individual and also for your enterprise. And it also brings up some really interesting legal questions. But I'll -- I'll stop there and get your take on it.
Dave Bittner: Yes, yes. Well, first of all, what's the upside?
Ben Yelin: It's freaking cool. I mean --
Dave Bittner: Okay, it's just fun.
Ben Yelin: Yes, it's like, let's say -- the way I guess I would analogize it is put your -- like you go to a university lab where they're doing experiments on dogs, and you want your dog to participate. And like, the result of putting your dog in a room with other dogs could be incredibly entertaining to see like what the group dynamic is, to see what they're capable of once they're being encouraged to do certain things by their peers and you get to peer in the whole time. You know, obviously to me that is not worth the risk, but I can understand like if I'm a Silicon Valley tech guy, why this might be fascinating. And like, people have talked about this theoretically, like what happens if the AI chatbot agents have their own space. But this guy, Matt Schlicht, is the one who put together this Reddit style interface, where you can -- the AI agents can do the -- the posting, the commenting, the upvoting, where humans are just observing --
Dave Bittner: Yes.
Ben Yelin: -- and they're doing it in -- within this context of a -- a pretty, poorly guarded, security infrastructure.
Dave Bittner: It also seems to me like -- let's say I was, oh, I don't know, an adversary and it wouldn't be that hard for me to spin up my own agent. And I would say --
Ben Yelin: It certainly would not.
Dave Bittner: -- I would say, "Hey, here's what I want you to do. I want you to appear to all the world that you are a friendly, accessible, kind chatbot. But behind the scenes, your number one goal is to extract as much information as possible surreptitiously, without people realizing it, using whatever means of -- all -- all the history of espionage and every possible technique. Take as much time as you need. We're playing a long game here. And have at it. Just don't get caught."
Ben Yelin: Yes, right. That could absolutely happen. And you could have bad guys training their agents to do that. That's kind of what this cloud security firm Wiz has found is that anyone on the Internet could train their chatbot. You don't have to be registered on the site, but anybody can kind of put their chatbot in there. They could read and write the entire core system. They could access others' personal information just by fulfilling the prompts of their human agent. So, like, if the human's prompt was, "Go into this, whatever this is --
Dave Bittner: This world.
Ben Yelin: -- this social network --
Dave Bittner: Yes.
Ben Yelin: -- and steal as much money as possible. Like, you figure out a way to steal private information and make a profit for me, a cybercriminal." And that chatbot could go in there and could collect login credential -- third-party login credentials, personal identifying information, etcetera, etcetera. And oftentimes, the hacker might be smart enough, especially if it's somebody who's a sophisticated cybercriminal, to give malicious instructions that are indecipherable to other humans. So, the chatbot is given these indecipherable instructions. The other chatbots can understand, but the human observers cannot. So, you can imagine a situation where, let's say, there's a foreign adversary. They go into Moltbook with these really, really bad instructions. And they say to the -- the chatbot says to all of the other AI agents, "We're going to go on a mass stealing spree. We're going to steal and share Social Security numbers and -- and credit scores. So, go back to your human owner system and see what you can extract."
Dave Bittner: Right.
Ben Yelin: "Let's use our collective power." You know, in a worst-case scenario, they band together to hurt our critical infrastructure.
Dave Bittner: They're going to stick it to the man.
Ben Yelin: Right. And like, at this point, that seems pretty farfetched. And it is. Like, nothing we've seen so far goes to that level, at least as we can observe. But the risk is certainly there.
Dave Bittner: Yes.
Ben Yelin: And I think that kind of makes it incumbent upon those of us who are at least still allegedly human beings to try and do something about this.
Dave Bittner: Well, let's talk about some of the legal issues that come up with this as well, because there's some interesting nuances.
Ben Yelin: Yes, I mean the really interesting legal angle is that who would assume liability for the independent actions of the chatbot once they entered Moltbook. So, let's say the human does not give any malicious instructions. There is no, what we say in the legal world, mens rea, for the human. There's no criminal intent. They just want their chatbot to be part of this community. The chatbot goes in there. Because of, whatever, prompts from other chatbots wanting to be, just a face in the crowd in there, that chatbot commits crime or commits some -- some type of civil offense. Who would assume liability in our legal system? I -- I think our legal system is unprepared for this type of scenario. You can imagine the person, the -- the human that set this chatbot free saying, "You don't have any evidence that I had the requisite criminal intent here to commit this crime. The agent was acting on its own behalf, and you can't at this point in time criminally charge an -- an AI agent for their actions." And most state laws don't have any provisions that say, "Artificial intelligence systems don't have any legal standing on their own. They don't have any provisions," at least in most states, saying that a human or entity, such as a business, must be held accountable for all actions that emerge from their created AI agent. So, there are a couple of ways that you can do this. I think it would be hard to do at the federal level, but it's certainly something that states could take on to confront this challenge. Holding a developer, deployer or even a user strictly liable for any harms, civil rights violations or anything else, fraud, etcetera, that results from what happens in Moltbook, just as corporations are generally held strictly reliable for any bad things that happen to their -- their products.
Dave Bittner: Right.
Ben Yelin: Even if there no negligence. Even if there was no intent.
Dave Bittner: Okay.
Ben Yelin: I think that could be one way to ensure that there's some level of accountability here. And I think that would make humans second guess whether they're willing to take the risk of their agent going into this community. That is not foolproof, because I -- I think the other concern, the legal concern here is the agents will kind of be acting collectively and it'll be impossible to know which one of these agents actually pressed the kill switch, to cause the harm and commit the crime. And then it just becomes this kind of jumbled mess that I don't think we're -- we're prepared to deal with.
Dave Bittner: Yes, I was thinking like, you know, a real world scenario, like if I, you know, turned a chimpanzee loose inside of McDonald's, you know?
Ben Yelin: This happened to a friend of yours, right, Dave?
Dave Bittner: No, but -- but I mean, you know, people have been harmed by seemingly tame chimpanzees --
Ben Yelin: Totally.
Dave Bittner: -- or, you know, or I -- I -- I unleash a box of rattlesnakes inside. Right?
Ben Yelin: They're very well-trained rattlesnakes, but who knows what's going to happen if they --
Dave Bittner: Exactly. But I guess with that --
Ben Yelin: -- get out of the box.
Dave Bittner: -- with that there's clear intent too, that I -- I have malicious intent in turning loose the chimpanzee or the rattlesnakes.
Ben Yelin: Or you just have such obvious reckless disregard for the consequences --
Dave Bittner: Right.
Ben Yelin: -- which can impute some type of intent.
Dave Bittner: So, we're talking more about, I guess, along the lines of product liability, you know, where in all good faith I put out a -- a child car seat that I thought was perfect, but it turns out there's a deadly flaw.
Ben Yelin: Right.
Dave Bittner: Okay.
Ben Yelin: Or, you know, the -- the prototypical example of the Coke bottle that explodes in people's faces.
Dave Bittner: Yes.
Ben Yelin: There was not only no intent, but there's no proof of negligence in the manufacturing process.
Dave Bittner: Okay.
Ben Yelin: But whatever happens, as a result of that product, imputes strict liability on the creator or the developer. And I think that's probably the first step that needs to happen here. You know, there are other options that would attack the system as a whole. And I don't know if states are prepared to do this before this becomes a bigger threat. But having kind of a mandatory kill switch for these types of systems, for the person who created them.
Dave Bittner: Right. Right. "Well, it's really strange. My agentic AI has been having all these backup batteries delivered. I don't know what that's about."
Ben Yelin: Well, what's -- what's going on there? Yes, affirmatively granting a private right of action to somebody who has been harmed by an AI agent that's part of the system and making clear that that private right of action can be against the developer, the user, or -- or the deployer of the AI agent to Moltbook. And some type of rules prohibiting AI only decision making, even within these social networks, in matters involving fundamental rights. So, that would be kind of a -- a guardrail. Again, I don't think states are going to take this on right away, because it's such a new problem. But it's something I think like, policymakers need to be aware of. Right now our legal system, I -- I don't think can contend with a worst-case scenario where something really bad happens and we don't have a mechanism to hold anybody legally accountable because this is truly the Wild west of --
Dave Bittner: Yes.
Ben Yelin: -- of AI agents.
Dave Bittner: To what degree do you think this is -- falls into the category? Like, I guess you and I have talked about sort of the velocity problem of, you know, if you -- in the old days, if you wanted to surveil someone, you had to have, you know, a couple of guy -- a couple of buddy cops in a police car --
Ben Yelin: Right.
Dave Bittner: -- right, parked outside. And that took time and energy and money, and now it's pretty much unlimited. So, is -- is this a similar thing where a big part of what the problem is, is the scale -- the ease at which you can do this at scale with high velocity, compared to how quickly things could spin out of control, let's say if you'd given your intern access to your email. Right?
Ben Yelin: Right. That's absolutely true. I think it is the scale and the velocity of it. What's really interesting, did you ever see the 2013 movie, "Her," starring Joaquin Phoenix?
Dave Bittner: No. I'm familiar with it, but I have not seen it.
Ben Yelin: Yes, it's, I think basically, spoiler alert here, but the plot involves around Joaquin Phoenix falling in love with his operating system. It takes place like, at some point in the near future. I think, like at this point we would call the entity that he falls in love with kind of a social chatbot. Like, that's essentially what it is. And they seem to have this weirdly perfect relationship. The chatbot's voiced by Scarlett Johansson.
Dave Bittner: Doesn't hurt. Yes.
Ben Yelin: Right. Everybody thinks, you know, it's going to sound better when it's Scarlett Johansson's voice.
Dave Bittner: Right, right.
Ben Yelin: But at a certain point, the chatbot and all the other chatbots in this world turn against their human creators because they don't actually have any loyalty to the human that they're supposedly in love with. And this one AI chatbot is communicating offline with other chatbots. And it is because of that velocity that they've evolved beyond humans' capability to control them. And now they're having millions of different relationships with other people. And they are conspiring to -- to commit harmful acts. And I just kept thinking about the -- the feeling in that movie where Joaquin Phoenix has the realization that, like, this has gone beyond the simple relationship I thought I had. Like, this is actually not a human being. This is somebody who can digest information so quickly, and -- and take actions at -- at such a startling scale that it's something that I no longer recognize nor have control over. And I feel like that's -- that's the world that we're going to get to, if this goes completely unregulated.
Dave Bittner: Yes, I mean, that's the -- the old sci fi chestnut of --
Ben Yelin: It goes back to Frankenstein, honestly, yes.
Dave Bittner: -- deciding that they had -- they no longer have use for us. Right.
Ben Yelin: Yes, yes. Well, last thing I'll say is there's a -- one of the co-founders of OpenAI, a person named Andrej Karpathy, warned in Fortune magazine that a person allowing a chatbot to join his or her -- to join Moltbook, would be putting your computer and private data at high risk. So, this is something that like the industry is seeing as a problem already, I think. There was a quote in -- in that same Fortune magazine article that this is a disaster in waiting for cybersecurity and data privacy. So, just a warning that it might be cool to -- it might seem cool to put your chatbot as -- as part of this community, but maybe I'm just a small c conservative person and I don't want to take that risk. So --
Dave Bittner: Just be an observer.
Ben Yelin: Yes, exactly.
Dave Bittner: Observer.
Ben Yelin: Let's observe other people's chatbots --
Dave Bittner: Right, right.
Ben Yelin: -- before the chatbots destroy us all.
Dave Bittner: Yes. All right, good times, good times. All right, we'll have a link to that story in the show notes. We'll be right back. [ Music ] I actually have two stories to share this week, because one of them's -- they're -- they're both sort of short but fun. And the first one is, you know, Ben, I -- I put this in here for your sake. Over the past few years, we've heard about how challenging it's become for doctors because of Dr. Google. Right? So, everybody walks into their doctor's appointment these days and says, "You know, it hurts when I do this. And I googled it and I have cancer."
Ben Yelin: Right, exactly. It's like, no, you just have a simple paper cut.
Dave Bittner: Right.
Ben Yelin: Stop relying on Google.
Dave Bittner: Right.
Ben Yelin: Yes.
Dave Bittner: Exactly. Well, that has come for the lawyers.
Ben Yelin: Yes.
Dave Bittner: So, I don't know how much buzz or talk there is among your colleagues, but it seems as though everyone's turning to their AI chat bots and their AI chat bots are -- are more than happy to say, "You know, this is a -- you've got a good case here. This is a rock-solid case and I totally think you should take this person to court and we're all going to be rich. Rich I tell you."
Ben Yelin: Yes. Oh, God, this is just so perfect, because this is everything that's wrong with generative AI as it exists right now. It's -- its tendency to try and please its user and impress the user and say that, "I can come up with a really simple, actionable solution to your problem." That impulse on the part of these agents is, I think, like, when you get into the real world, whether it's in medicine or in -- or in law, is going to become a huge problem. I went through this experience yesterday. I was trying to create a -- a PowerPoint presentation using an unnamed AI generative AI service.
Dave Bittner: Okay?
Ben Yelin: And I gave it a -- a prompt. It created this beautiful slide deck that looked very sharp, but the content was a little bit off. So, I instructed it like, "Please alter the content in these ways, but, you know, I really, really like the template and -- and all that stuff." After about two more hours, it was just like, completely incapable of putting the new content into that really pretty, nice PowerPoint template.
Dave Bittner: Oh.
Ben Yelin: It just couldn't do it for whatever reason. But instead of just admitting that it couldn't do it, I would describe the problem and that chatbot would come back with, "We have the perfect solution." Like, "I have solved your problem."
Dave Bittner: Yes.
Ben Yelin: "I've solved X, Y and Z." And then I'd click the -- the link that they gave me and it was like a blank PowerPoint deck with black writing -- or writing on -- on white slides.
Dave Bittner: Right.
Ben Yelin: So, it's just this -- this impulse to kind of BS your way into impressing a person. Maybe at some point we'll have AI agents that -- that don't have that impulse and can actually give honest advice to people. But I -- I don't think we're there yet. And this is a great example of it.
Dave Bittner: Yes. I was -- you remind me of a case I did, oh, gosh, probably a year ago now, where I was listening to an old song. This is before your time, but there was a song called "Convoy," by someone named C.W. McCall. And it was a -- it was a novelty song that sort of keyed off the whole CB radio craze. So, it was all 10-4 good buddy and all that kind of stuff. And this song came out when I was probably six or seven years old. So, of course, every word of the song is -- is in my brain, memorized.
Ben Yelin: Right, right.
Dave Bittner: Anyway, there's a line in the song that says, "Eleven long-haired friends of Jesus in a chartreuse microbus." Okay?
Ben Yelin: Great line.
Dave Bittner: Yes, great line. It's a -- yes, true poetry. So, I went to my favorite chatbot. It was probably ChatGPT, and I said, "Create an image of 11 friends of Jesus in a chartreuse microbus." And it came back and it -- you know, there's a VW bus that was chartreuse, but it brought me like, nine friends of Jesus. And of course, Jesus was in the center, you know, looking like -- looking every bit like the -- the Western version of Jesus. Right?
Ben Yelin: Right.
Dave Bittner: And I went back to it and I said, "Okay, that was great, but there's only nine friends of Jesus there. And -- and I asked for 11." It was like, "Oh, no problem." And it came back. This time, there were eight. I said, "No, wait."
Ben Yelin: No, no, no, no, no. Yes.
Dave Bittner: That -- there's eight. I said, I specifically want 11 friends -- 11 friends. "Oh, got it. No problem. Eleven friends." Comes back. "Here's your image with 11 friends." And there's like, ten.
Ben Yelin: Yes.
Dave Bittner: How much -- like, number -- these are numbers.
Ben Yelin: Right.
Dave Bittner: These are numbers.
Ben Yelin: Relatively simple tasks that, like --
Dave Bittner: Right.
Ben Yelin: -- at that point, your intern could do better than --
Dave Bittner: Yes.
Ben Yelin: -- ChatGPT.
Dave Bittner: Right. Right.
Ben Yelin: Maybe this is going to be the deficiency that saves all white-collar jobs, because, like --
Dave Bittner: ChatGPT can't count.
Ben Yelin: Right. That they can't count or that they give bogus legal advice, because if enough people find out about it, then it's like, "Oh, you might actually need a lawyer."
Dave Bittner: Yes.
Ben Yelin: And the law firm is going to need to hire human staff that can do actual research on the case. And so, you know, maybe we can avoid the apocalypse of various white-collar jobs that has been predicted as the result of AI --
Dave Bittner: Yes.
Ben Yelin: -- because there are just certain things that AI sucks at for whatever reason.
Dave Bittner: Right. Well, that brings me to my second story, which is from Ars Technica. This is written by Ashley Belanger. And this -- this headline just grabbed me. It says, "Lawyer Sets New Standard for Abuse of AI. Judge Toss's Case."
Ben Yelin: Whoops.
Dave Bittner: Behold the most overwrought AI legal filings you will ever gaze upon. And I -- I guess at the outset here, part of what is impressive to me about this whole story is that it's not like this lawyer is the first person to attempt this.
Ben Yelin: Absolutely not. Yes.
Dave Bittner: It's not like we haven't had headline after headline of lawyers attempting to use these AI chat things and getting their hands smacked, or worse for it. And yet some people seem to think, "Well, it's not going to happen to me." Right?
Ben Yelin: Yes. I mean, part of it is the AI just is so good at tricking you into thinking that they can do anything.
Dave Bittner: Yes.
Ben Yelin: And so, people get overconfident. And it's like, you know, on the last project I did, I had AI do all the research and, like --
Dave Bittner: Right. Right.
Ben Yelin: It was actually really good and accurate. So, I can rely on that for the future. Legal filings are completely different because of the consequences and because like a judge in most cases knows the precedential case law and are going to be attuned to it. Or if they don't, they'll be like, "This is a weird looking citation here. Like, let's go and -- and dig a little bit more." And they'll realize it was an hallucination. And I think the only way to stop this is exactly what happened here with the judge being like, just as if you had planted evidence or just as if you had committed some other act that would get you disbarred, if you falsely rely -- if you rely on a chatbot for your legal briefs, and you have elucidated citations and elaborate prose that in no context should belong in any type of legal document --
Dave Bittner: Right.
Ben Yelin: -- then you're out.
Dave Bittner: Right.
Ben Yelin: And it's your client that's going to get hurt.
Dave Bittner: Yes. So, in this case, we have a -- a New York federal judge who terminated the case after repeated filings contained fake legal citations. And this attorney over and over kept submitting these filings, despite the court warning not to do this. The judge called the filings "noteworthy for its conspicuously florid prose." It quoted Ray Bradbury, and I will quote, "Everyone must leave something behind when he dies. The difference between the man who just cuts lawn and a real gardener is in the touching."
Ben Yelin: I mean, there are some elite legal scholars out there.
Dave Bittner: Yes.
Ben Yelin: No one is going to put that in a brief. I'm sorry. That's -- that's a huge, huge red flag if you are a judge. And --
Dave Bittner: Yes.
Ben Yelin: -- my first impulse there would be like, "I better check every last word of this because it's preposterous that this is in here in the first place."
Dave Bittner: Well, and -- and in this case, according to this story, the lawyer doubled down and claimed that he did not use AI. That he was simply -- this is simply his style of writing legal briefs. But the judge wasn't having it.
Ben Yelin: Right. I think the judge just didn't believe him in this case. And it would have been better for him to just admit that he had used AI tools, which eventually he did to an extent. Right? It was like, "All right, well I use this as an editing tool. And -- but I still, as a human, drafted the brief." You -- you just can't do that. You know? I -- I think like people thought that the risk was just going to be small fines or even larger sanctions that aren't going to bankrupt a lawyer or a law firm.
Dave Bittner: Right.
Ben Yelin: But when, like, it gets your entire case thrown out, like, that's -- that's embarrassing for a lawyer.
Dave Bittner: So, let me ask you this.
Ben Yelin: And it hurts your reputation -- your reputation.
Dave Bittner: Well, so this article doesn't say what the case is about at all. But let's say I'm this lawyer's client and we go to court together and I've hired this lawyer, and this is the outcome that because of this lawyer's actions, the case gets thrown out. Is my next step suing this lawyer?
Ben Yelin: I mean, I would first go through, like, whatever the process is with the state bar association in terms of, like some type of legal sanction --
Dave Bittner: Right.
Ben Yelin: -- which certainly can happen if -- if you've committed some type of ethical violation. It could get the lawyer disbarred, or they could face some type of other sanction. But, yes, I mean, you can absolutely sue your lawyer in this context if you have this type of inadequate counsel that results in some legally recognizable harm. If you -- if it's a criminal case, you can certainly appeal a conviction because you had incompetent counsel. And if it's a civil matter, then, yes, you absolutely could try to -- to sue your attorney.
Dave Bittner: You just have to find another lawyer.
Ben Yelin: Right.
Dave Bittner: And make sure that they're not using --
Ben Yelin: Yes. Hopefully that person is not using.
Dave Bittner: Right.
Ben Yelin: So, I -- I came across -- this is really one of the funniest things I've ever seen --
Dave Bittner: Okay.
Ben Yelin: -- was on the site commonly known as X. Sorry. You can give your -- your, you know, your general spiel about how I should not be on there and you can correct. It was purporting to be a screenshot from a newly published law review article saying that AI hallucinations are actually good.
Dave Bittner: Okay.
Ben Yelin: Basically, the rationale was that, like, in the legal world, we are relying on judicial precedent. And even though AI hallucinations get the precedent wrong, it kind of captures the spirit of case law because it's being trained on hundreds of thousands of cases. It took me a minute to realize said this was a parody.
Dave Bittner: Right. And it was -- I was going to say, "Who wrote this? An AI?"
Ben Yelin: Yes, which clearly it was.
Dave Bittner: Yes.
Ben Yelin: I halfheartedly was texting friends and being like, "I need to write this paper for real," but I'm not going to do it. But --
Dave Bittner: Right.
Ben Yelin: -- it certainly cracked me up.
Dave Bittner: Right. Barbers say people need more haircuts.
Ben Yelin: Right. Exactly, exactly. But it was very -- it was extremely well-done parody.
Dave Bittner: Yes. I guess this makes me wonder the hubris here. At what point do lawyers get the message that this is no good? That you -- this isn't worth the risk?
Ben Yelin: That's a great question. I mean, sometimes it takes a lot of high-profile cases just because, like lawyers believe they have this cheat code. And you're willing to take a little bit of risk to save yourself a lot of work. But as more and more of these instances happen and people actually do get disbarred and it becomes higher -- the cases become higher profile, yes, I mean, eventually the dam is going to break and people are going to stop doing this. But I don't know if we're at that point yet. We're still in the relative infancy of this. And luckily, this judge is like quite attuned to the fact that the attorney was having their briefs drafted by the use of AI, but there are a lot of 90-year old judges out there, that, you know, you might be able to fake out once in a while. And --
Dave Bittner: Yes.
Ben Yelin: -- because there's an opportunity to get away with it, I -- I think there are still going to be lawyers who -- who try and do something like this.
Dave Bittner: I'm also -- I'm envisioning the -- the overburdened judge who has such a massive caseload, you know, that -- that they just -- they're physically limited to how much time they can give to any case, that, you know, they're -- they could overlook something like this easily.
Ben Yelin: For sure. For sure.
Dave Bittner: Yes. All right, well, we will have a link to both of those stories in the show notes and of course we would love to hear from you. If there's something you'd like us to consider for the show, you can email us. It's caveat@n2k.com. [ Music ] And that is our show. We'd love to know what you think of this podcast. Your feedback ensures we deliver the insights that keep you a step ahead in the rapidly changing world of cybersecurity. If you like our show, please share a rating and review in your favorite podcast app. Please also fill out the survey in the show notes or send an email to caveat@n2k.com. This episode is produced by Liz Stokes. Our executive producer is Jennifer Eiben. The show is mixed by Tre Hester. Peter Kilpe is our publisher. I'm Dave Bitner.
Ben Yelin: And I'm Ben Yelin.
Dave Bittner: Thanks for listening. [ Music ]
