This weekly roundup thread is intended for all culture war posts. 'Culture war' is vaguely defined, but it basically means controversial issues that fall along set tribal lines. Arguments over culture war issues generate a lot of heat and little light, and few deeply entrenched people ever change their minds. This thread is for voicing opinions and analyzing the state of the discussion while trying to optimize for light over heat.
Optimistically, we think that engaging with people you disagree with is worth your time, and so is being nice! Pessimistically, there are many dynamics that can lead discussions on Culture War topics to become unproductive. There's a human tendency to divide along tribal lines, praising your ingroup and vilifying your outgroup - and if you think you find it easy to criticize your ingroup, then it may be that your outgroup is not who you think it is. Extremists with opposing positions can feed off each other, highlighting each other's worst points to justify their own angry rhetoric, which becomes in turn a new example of bad behavior for the other side to highlight.
We would like to avoid these negative dynamics. Accordingly, we ask that you do not use this thread for waging the Culture War. Examples of waging the Culture War:
-
Shaming.
-
Attempting to 'build consensus' or enforce ideological conformity.
-
Making sweeping generalizations to vilify a group you dislike.
-
Recruiting for a cause.
-
Posting links that could be summarized as 'Boo outgroup!' Basically, if your content is 'Can you believe what Those People did this week?' then you should either refrain from posting, or do some very patient work to contextualize and/or steel-man the relevant viewpoint.
In general, you should argue to understand, not to win. This thread is not territory to be claimed by one group or another; indeed, the aim is to have many different viewpoints represented here. Thus, we also ask that you follow some guidelines:
-
Speak plainly. Avoid sarcasm and mockery. When disagreeing with someone, state your objections explicitly.
-
Be as precise and charitable as you can. Don't paraphrase unflatteringly.
-
Don't imply that someone said something they did not say, even if you think it follows from what they said.
-
Write like everyone is reading and you want them to be included in the discussion.
On an ad hoc basis, the mods will try to compile a list of the best posts/comments from the previous week, posted in Quality Contribution threads and archived at /r/TheThread. You may nominate a comment for this list by clicking on 'report' at the bottom of the post and typing 'Actually a quality contribution' as the report reason.
Jump in the discussion.
No email address required.
Notes -
New case law just dropped[^1]: a guy was charged with a $300M securities fraud. Before his arrest he used Claude (the consumer product) to research his own legal situation. He then handed the outputs to his defense counsel and claimed attorney-client privilege. The prosecutor said "no, that's not how this works, that's not how any of this works", and the judge agreed[^2]. That means that as of this decision, precedent says that if you go to chatgpt dot com and say "hey chatgpt, give me some" legal advice, that's not covered under attorney-client privilege.
On the one hand, duh. On the other hand, it really feels like there should be a way to use LLMs as part of the process of scalably getting legal advice from an actual attorney while retaining attorney-client privilege.
I expect there's an enormous market for "chat with an AI in a way that preserves attorney-client privilege", and as far as I can tell it doesn't exist.
It was also interesting to read the specific reasoning given for why attorney-client privilege was not in play:
I notice that none of these reasons are "conversations with AI are never covered by attorney-client privilege." They're all mechanical reasons why this particular way of using an AI doesn't qualify. Specifically:
The prosecutor also argues that feeding what your attorney told you into your personal Claude instance waives attorney-client privilege on those communications too. If a court were to agree with that theory, it would mean that asking your LLM of choice "explain to me what my lawyer is saying" is not protected by default under attorney-client privilege. That would be a really scary precedent.[^4]
Anyway, I expect there's a significant market for "ask legal questions to an LLM in a way that is covered by attorney-client privilege", so the obvious questions I had at this point were:
For question 1, I think the answer is "no" - a cursory google search[^5] mostly shows SEO spam from
So then the question is "why doesn't this exist" - it seems like it should be buildable. Engineering-wise it is pretty trivial. It's not quite "vibe code it in a weekend" level, but it's not much beyond that either.
After some back-and-forth with Claude, I am under the impression that the binding constraints are
None of these seem insurmountable to me. I'm picturing a workflow like
Anyone with a legal background want to chime in about whether this is a thing which could exist? (cc @faceh in particular, my mental model of you has both interest and expertise in this topic)
[^1]: [United States v. Heppner, No. 25 Cr. 503 (JSR) (S.D.N.Y. Feb. 6, 2026)] (https://storage.courtlistener.com/recap/gov.uscourts.nysd.652138/gov.uscourts.nysd.652138.22.0.pdf). The motion is well-written and worth reading in full. [^2]: Ruled from the bench on Feb 10, 2026: "I'm not seeing remotely any basis for any claim of attorney-client privilege." No written opinion yet.
[^3]: This argument feels flimsy, since attorneys send privileged communications through Gmail every day, and Google can and regularly does access email content server-side for reasons other than directly complying with a subpoena (e.g. for spam detection). It could be that the bit in Anthropic's TOS which says that they may train on or voluntarily disclose your chat contents to government authorities is load-bearing, which might mean that Claude could only be used for this product under the commercial terms, which don't allow training on or voluntary disclosure of customer data. I'm not sure how much weight this particular leg even carried, since Rakoff's bench ruling seems to have leaned harder on "Claude isn't your attorney."
[^4]: A cursory search didn't tell me whether the judge specifically endorsed this theory in the bench ruling. So I don't know if it is a very scary precedent, or just would be a really scary precedent.
[^5]: This may be a skill issue - I am not confident that my search would have uncovered anything even if it existed, because every search term I tried was drowned in SEO spam.
Don'y you have something like Litigation Privilege in the US where general communications made for the primary purpose of assistance with ongoing or reasonably contemplated (a wide net) litigation is protected regardless of who the parties in the conversation are? That would solve the whole problem.
Alternatively: use an open source local model. They are getting good enough these days for litigation tasks.
More options
Context Copy link
Law noob here. If you reveal the existence of evidence to your lawyer in an email, even though the email is protected, is the evidence still subject to discovery? E.g. If I emailed my lawyer that I told Amy that I committed a crime, could that be used as evidence to compel Amy to testify, even if there was no other justification?
Everything I’ve learned from Hillary/Epstein/law in general is to never send emails ever. If email didn’t exist nobody would ever get convicted of anything.
More options
Context Copy link
I'm not sure what the point of this would be. If you're represented by a lawyer, why are you asking an LLM questions about your case. That's the point of having a lawyer! If the lawyer can't explain things sufficiently, then you simply don't have a good lawyer. In any event, this wouldn't work because you would still be disclosing the information to a third party. The email question is complicated, but it ultimately isn't comparable. For starters, we have to be very careful with our email communications in general because there is a lot of confidential information (including information that would be shared with the opposing party anyway) that could get leaked—social security numbers, financial information, medical information, proprietary business information, etc. Any law firm is going to use a secure email server run by their IT company. Any email sent from the attorney's account is going to be encrypted, and they should ensure that any email the client sends will be encrypted as well.
If for some reasons these precautions aren't taken, there's not gotcha that I'm aware of that says because something hidden in Google's TOS that says they can read your emails would make communications from an unsophisticated party to an attorney unprotected because they were shared with a third party. When it comes to attorney communications, most judges would probably say that the sender had a reasonable expectation of privacy in email communications and didn't expect that the server host would have access, especially a large host like Google that deals with millions of emails an hour. The situation with LLMs is different because you're deliberately giving the third party company the information with the expectation that they will provide an output, not just using them as a messenger or having them store it for you.
When I was doing consumer-side stuff, LLMs weren't really a thing yet, but Google definitely was. I told clients that if they had questions for me to write them down and call me after enough time had passed that they didn't anticipate having any more (usually a couple weeks). I expressly warned them not to start Googling for answers. The reason for this isn't because I was protecting billable hours (I was charging flat fees and it was theoretically costing me money to take their calls), but because I didn't want them scaring the shit out of themselves. There's a lot of bad or inapplicable legal information online, and it's easy for someone to get worked up based on information that's irrelevant. If they really wanted more information I'd tell them to check a NOLO guide out of the library because at least I could vouch for the accuracy of the information, but told them to keep in mind that I paid hundreds of dollars a year for practice manuals that go into much more technical detail than anything they're going to find and am also familiar with local practice. In other words, if they get concerned that I didn't take some factor into consideration then rest assured that if they're reading about it on the internet than I did. There were, of course, a few people who felt the need to argue with me and any time I'd set them straight they'd go back to Google to prove me wrong, but they were just assholes.
Because if you ask a question you get 30 seconds to a minute of explanation. The lawyer, on the other hand, gets (at least) an hour billable. For lawyers willing to do business with scrubs like me (vice lawyers and firms that specialize in multi-million or billion dollar business only--God only knows how much they cost), that's $600-800 for something I could have gotten for free if I'd only taken the time to type out the full context. Obviously, this could go really sideways really quickly and when life, liberty and the pursuit of happiness are on the line, that's not good. But I'd like to have it as an option for the times it feels safe.
More options
Context Copy link
Few lawyers are particularly good at explaining matters in nontechnical terms, and perhaps more immediately, lawyers have limited hours in a day and expect compensation for the time they do offer. Maybe in the case of the 300m embezzlement that's a bad place to save
pennieshundred-dollar-bills, though depending on the facts of the case that might not actually be 300m in this guy's pockets. But for the average person, an extra consult with an entry-level lawyer can cost them a day's pay.At minimum, you want to go into that prepared. At the more extreme end, if you're being charged or sued as a rando, it's suddenly the most important part of your day.
I agree that there are bad attorneys out there, I'm just skeptical that self-help remedies offer any improvement. I did consumer-side law for a few years and the amount of "preparation" a client does in a legal sense (and not just having the right documents together) is usually inversely proportional to how easy it was for me to handle their case. At best these people would read about worst case scenarios and freak themselves out; these were the easy cases because these people tended to be very trusting and did an emotional 180 at the end of the consult. The worst were the people who were convinced they knew how I should handle the case and second-guessed my every decision. No, I am not going to try that One Weird Trick that's not going to work and might get me sanctioned. Yes, after much cajoling, I may be willing to put together a bunch of exotic entities that you heard about, but only after trying to talk you out of them, and only because if I lose you as a client you'll go to someone else who will charge you more and won't do the job right.
I did a lot of bankruptcy and estate administration and the problem with those isn't so much that clients have a ton of questions but that there's a flurry of activity right at the beginning followed by a lot of waiting, and they get worried if they haven't heard from you in a while and all you can tell them is that it's a waiting game. I'm not going to get into the workflow and how the consults are usually constructed, but in these kinds of cases the client is meeting with the attorney enough in the early stages that there's plenty of opportunity to ask questions. Like I said, LLMs weren't a thing quite yet, but there was a lot of Google, plus people like Suze Orman and Dave Ramsey, and I specifically instructed my clients to avoid the internet and pop financial gurus and if they wanted general information I recommended good publications that were easily available in libraries. But I can't imagine that telling them to direct their questions to an LLM would inspire much confidence.
More options
Context Copy link
More options
Context Copy link
I think "don't have a good lawyer" is pretty common. Probaly about as common as "don't have a good doctor." And to most people who aren't themselves in the field the skill level of these professionals is quite illegible.
I'll share a personal anecodte. About a decade before ChatGPT's launch in a few minutes of Googling I was able to shield my deceased grandmother's house from being seized by the state. She had made use of in-home healthcare services from the state and this gave them a claim on the house after her death. She'd received years of care and the property was a two bedroom condo in a low cost area, so they would've gotten the whole thing. But, the statute included clear exemptions for if any of her surviving children were blind or otherwise disabled, and both of her living children did.
I called the state office that would've pursued recovery, they confirmed the rules, and said that they wouldn't pursue recovery in that case.
The probate lawyer that she had pre-arranged had never heard of such an exemption. If I didn't check on it myself then that property would've been the state's.
I just posed the scenario to ChatGPT and it was able to cite the rules in which blind or permanently disabled children will block recovery.
I agree that there are plenty of bad attorneys out there, but the situation you describe is malpractice.
"Potential malpractice" is a very good reason to do your own research, whether that be Googling or LLMs.
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
I expect many people do not have a good lawyer, by this standard. Or can't afford a good lawyer, for the number of hours of the lawyer's time they'd need to actually understand what's going on with their case and what that means in practical terms for them.
To your point, this is probably the real answer. If the logs of the chat don't exist, they're not going to show up in court.
More options
Context Copy link
My clients love doing their own "research" and trying to "help" with the case. I'm just some public pretender, so an LLM/their cellmate/their cousin who knows a guy who got his case dismissed by filing a sovereign citizen motion all know way more about the law than I do.
More options
Context Copy link
More options
Context Copy link
I'm looking at the linked court listener docket and I think your description is a little misleading. What you linked to is a motion by the United States as to why the documents shouldn't be covered by attorney client privilege but I don't actually see a ruling by the judge granting or rejecting the motion. So there has not yet been a decision in this case as to whether the documents are privileged, just an argument by the prosecution that they shouldn't be.Missed the minute order on 2/10.That said, I think the prosecution is basically correct. Imagine rather than an AI you email a friend asking their non-attorney legal advice. Maybe you discuss statutes and case law or possible defenses. Would those emails be privileged? What if you fire up your search engine and start searching for statutes you may have violated. Relevant case law. Defenses. Is the existence or content of those searches privileged? My intuition is that they would not be. I don't see what is different about AI such that its use generates attorney-client privilege but the use of other legal research tools or avenues does not.
I suppose I think the obvious way to get AI input and also remain privileged would be to use a lawyer as a kind of middle-tier. User query -> lawyer -> AI. AI response -> lawyer -> user.
I don't know much about law but this seems very silly. Should the user really be at a disadvantage because he is not representing himself? Why should the lawyer's actions be beyond reproach, but not the user's, unless the user says the magic words "I'm representing myself" and then 5 minutes after that re-hires their lawyer?
Anything tangentially related to the legal battle should be inadmissible as evidence.
Why do we want to extend privilege, which is a situation where we all pretend that what we know ain't so, to LLM queries?
An LLM is a fancy wikipedia being spewed back at you after being re-jiggered by a fancy matrix of polynomials. I don't see why chats with an LLM bot SHOULD be admissible.
Agreed. LLM conversations should not be admissible for the same reason that a schedule of events that you make on Google Sheets to help you track the events of your case and save on Google Drive isn't admissible, even though it's a conversation with a third party. The US discovery process is absolutely cooked if you ask me, no sensible person would come up with what they have if starting from scratch.
More options
Context Copy link
Why wouldn't it be? Notes to oneself aren't privileged. If I keep a notebook it isn't privileged. If I ask you a question it isn't privileged. My lawyer, my priest, and my wife are privileged; we do so to protect those certain relationships. Why do we want to extend this to LLMs?
What you're looking for, I guess, is something like why would it be probative to talk about someone's LLM history? Would it be more prejudicial than probative to admit them as evidence? But that's very different from privilege.
It's not just about relationships. The lawyer role itself is highly privileged--your lawyer can google anything, say anything, etc. and it's not admissible as evidence. It's not [communications with certain people] that are privileged, it's [virtually everything your lawyer does, plus communications with a couple other people] that is.
I think this is weird and bad policy. The client should be capable of doing a little bit of his own legal work.
So do you think that all google searches, internet queries, LLM queries should be inadmissible in court? Or only those that can plausibly be called legal research?
I think whatever protections we apply when someone represents themselves should also apply when they hire a lawyer. Direct legal research into a case is already protected. Slightly-indirect legal research into "does my lawyer know what they're talking about" should also be protected.
In practice it really is a priesthood. Given judicial prejudice I doubt an extremely competent lawyer at the top of his game would get half as good an outcome representing himself as he would get by hiring a bottom-of-the-barrel lawyer. So appealing to rules as written and drawing a comparison is doomed from the start since it's already not how courts function in reality at all.
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
First, I am amazed by the selective stupidity of people. A teenager getting sued by the RIAA (do they still exist?) for downloading a copyrighted mp3-file going to the big LLMs for advice is something I can see. I have no idea how to commit a 300M$ securities fraud in the first place, but if I did I would probably find a spare couple of thousands to discuss my troubles with a lawyer in person.
Also, this is a priesthood ruling that entities who are not ordained priests are not allowed to function as priests. Zero surprise there.
Your workaround would rely on straw lawyers who just start the chat and leave answering the questions to the LLM. The problem with that approach is that it makes the lawyer liable for all the answers the LLM gives you. After all, if you are in a privileged discussion of your legal woes with your lawyer, and a third party opines that if you just confess to everything, you will not be punished, you would reasonably expect your lawyer to refute that claim, and might sue them if they did not.
The same incentives which makes court not recognize LLM communication as privileged even if it covers topics traditionally covered by lawyers would also make them go after any lawyers who start a LLM session and then do not verify the responses.
I think that this is a case where a technical solution is much more apt to solve the problem than a legal one -- just as it is much easier to encrypt your communication than to prevent the NSA from snooping on it.
In most criminal cases, the FBI has no backdoor to the devices of the suspect (and if the NSA does they would not be willing for that fact to become common knowledge just to convict some fraudster). Also, it seems unlikely that they will have hit random IT service providers with an order to record all communication of a suspect. (NSLs do not cover content, but I guess a judge could likewise force a party to record and gag them about that fact.)
Much more likely is that you and the LLM provider will be hit with a subpoena for recorded information. Thus it is sufficient to ensure that there is no record of your conversation. So you want a LLM provider which does not keep records without a court order, which is probably something you can get at enterprise level. And then you simply buy a thumb drive, install Ubuntu (or whatever) on it, boot it, have your little discussion, turn off your PC and microwave the drive. (Sending a transcript to your meatbag lawyer is riskier but not tremendously so, my understanding is that the courts generally do not snoop on lawyer communication in case there is something not covered by client-attorney-privilege. Just do not use your normal mail account!)
Of course, the seriously paranoid will want to run an open-weight LLM locally instead, or at least use a Chinese one whose operators are much less likely to cooperate with US authorities.
The UK bar does something very similar. If you're a company being sued in the high court it's a criminal offense + contempt of court if you don't hire a barrister to represent you, regardless of your size or financial position. Fortunately you can recover costs if you win, but they are rarely 100% of the actual costs (more like 60-70%).
Apart from the split profession in England*, which doesn't actually add much to the cost because junior barristers are cheaper than experienced litigation solicitors, this is true basically everywhere. A company can't self-represent (it has no mouth to speak in court and no hands to sign documents) and non-lawyer employees are banned from representing a company almost everywhere (for "grownup Court" this includes England, Scotland, and AFAIK all 50 US States). The English system at least allows non-lawyer employees to represent their corporate employer in small claims court, which not every US state does.
* Solicitors do general legal work, and are not usually allowed to argue in the courts that handle high-value civil cases and serious crimes. Barristers do trial work, and are prohibited from doing certain types of administrative work in a way which means only a sophisticated client can hire one directly. The normal way you litigate is to hire solicitors who will handle pre-litigation correspondence, file the case, deal with discovery and settlement negotiations, and hire a barrister for you to handle trial prep and the trial itself once it becomes clear that the case will probably go to trial.
I feel there should probably be an exception for a guy appearing on behalf of a company when he's sole owner of that company.
It depends on the state but there usually is an exception depending on the circumstances. In PA owners of sole proprietorships and privately-held LLCs can represent themselves in small claims court, where the jurisdictional limit is $12,000 and procedure is more relaxed. Once you get to big boy court things are dramatically different. Judges aren't going to let you slide on deadlines and procedural errors. (It's one thing to be unable to meet a deadline or file something wrong, but you have to get permission from the court or opposing counsel and rectify any errors as soon as they're discovered. Opposing counsel is often willing to cut you some slack since lawyers in a city deal with one another all the time and want to be extended the same courtesy. A pro-se litigant isn't going to have that expectation.) It's really easy to fuck things up, and that's before you even get into the legal arguments. You might as well make a rule that they can save themselves the trouble and go straight to remedies.
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
It's pretty obvious from the context here that this guy wasn't just trying to get background information or looking up the definition of "included offense"; if that were the case it's unlikely his attorney would even contest handing over the documents. He was probably laying out what he did in detail and trying to see if the LLM's advice corresponded to what his attorney was telling him, or doing something else that required him to disclose incriminating information.
More options
Context Copy link
More options
Context Copy link
Can anyone explain why the defense council didn't held his mouth shut? Also what damning thing could be in documents about hypothetical situation. I mean obviously the results of AI prompts are inadmissible - too much hallucinations
Here in the UK at least defense counsel would breach their own professional duties to the court if they were challenged on the existence of such documents and ended up giving the impression (in any way whatsoever) that such documents didn't exist when defense counsel knew they existed. They don't have to volunteer over the fact that they exist but once identified must respond in a fully truthful manner. It's a good thing that keeps parties honest overall I think.
More options
Context Copy link
IME, the angle here is less about the content of the AI's responses than the user's questions. You sometimes see indictments that include, say, search terms a defendant entered around the time they committed their crimes as evidence for consciousness of guilt. They knew they were doing something that was (or could be) illegal.
More options
Context Copy link
I believe defense counsel needed to identify the documents to assert they were privileged. Otherwise they would be presumptively non-privileged items in the defendant's possession and automatically subject to examination by the prosecution.
More options
Context Copy link
More options
Context Copy link
In another intersection of law and LLMs, a study whose was published this month about how do LLMs compare to federal judges in terms of following correctly impacting their ruling:
LLMs outperformed human judges.
This is exactly what I'd expect. Humans are full of bias. For example, the hungry judge effect (judges are more likely to give harsher sentence just before lunch).
Daniel Lakens: Impossibly Hungry Judges
Andreas Glöckner: The irrational hungry judge effect revisited: Simulations reveal that the magnitude of the effect is overestimated
In short: the supposed effect is absurdly large, with the probability of a favorable ruling going from 65% to almost 0% before a break. The far more likely explanation is that, since the order of cases is not random, worse cases are scheduled last. In particular, it makes sense for judges to put short cases last rather than ones anticipated to go over time, and losing cases are shorter.
This sort of explanation should be the first thing we consider when hearing about a supposed effect like this. Outside of actual randomized control trials selection bias tends to be more powerful than the effect being investigated and common methods like "we controlled for some things we thought of and assumed any remaining discrepancy was the effect we're looking for" are inadequate for dealing with it.
More options
Context Copy link
IIRC there's some question about if the results of that study were due to other cofounding factors, for instance judges scheduling "tougher" cases earlier in the day.
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
I'm not a lawyer, but your proposed workflow sounds a lot like "CC a lawyer on an email conversation that you don't actually need to involve a lawyer in, solely so that you can claim attorney-client privilege on the whole conversation". That's something that big companies like Google have tried, but judges don't look very kindly on it: https://www.reuters.com/technology/landmark-google-ruling-warning-companies-about-preserving-evidence-2024-08-06/
As far as I can tell
I agree with your expectation that judges would not like that product very much.
More options
Context Copy link
More options
Context Copy link
I asked an LLM about this and it pointed out that the judge dismisses that work product doctrine applies here:
And that work product doctrine is governed by rule of civil procedure 26 (b) (3):
"by or for another party" (the other party being the one subject to discovery) clearly includes the party that is the target of the litigation.
The judge doesn't appear to cite anything that contradicts this. He thinks he does, with this bit from In Re Grand Jury Subpoenas:
But that bit doesn't establish that work product applies exclusively to the attorney's work products. Here's the full passage:
The judge's quote cuts off a crucial "and," which establishes that regular documents that aren't litigation prep that wind up in the attorney's posessesion aren't automatically work product. So the client's conspiracy laid out in Excel that he also transmitted to his lawyer is not work product. But his queries about how much trouble he might be in and what he can do to shield himself very well may be.
More options
Context Copy link
Maybe the actual product here is something like proton mail (I think that was the one). E.g. this AI doesn't keep logs so there's nothing to subpoena (tripwire disclaimer, or host in a friendly jurisdiction). Or we could go for a robot priest, Confessional AI, only the machine god can absolve you of your sins.
Does seem like there's some room for nuance here. Lawyers communicate via a variety of means, e-mail, etc. (though a lot do seem to use a specific messaging services), but I would expect assistive technologies would be protected (screen reader, interpreter, translator, what about that old phone service for deaf people, I believe that was not subpoenable) and to the extent one could do a service that helps you interpret the legalese from your lawyer how much does it differ from those?
More options
Context Copy link
LLMs as legal advise is one of the largest fails so far. They are simply extremely bad at the job. Even big corps like Westlaw and Lexis that have been trying it out have seen poor results. The fact is there is a gap in tech.
But this person obviously should have known better. How long has it been since the first person who googled "how to dispose of a dead body" had that introduced at trial? 2 decades? 3? "AI" at this point is just google with a little extra kick.
People have called me old-fashioned for this, but I still think that legal digests are the best way to conduct research since you can browse cases broadly by category and read thumbnail descriptions rather than have to dive into the case itself. Unfortunately I don't have a law library at work, so I have to make do with Lexis. Luckily I don't have to do research very often.
With LLMs, I think there's a general problem whereby people who aren't in a field and don't know what people in the field actually do all day confidently assert that some piece of technology will make them obsolete. Lexis or Westlaw could develop a perfect research tool that gave me all the relevant material on the first try every time, and it might save me a few hours per year. And even at that, a lot of legal argument involves a kind of inferential knowledge that you aren't going to find explicitly stated in caselaw. Just for fun, I constructed a simple scenario loosely based on an argument that took place last week:
In the real case there were additional factors at play that complicated the situation, but this distills a basic question. I won't reproduce the overlong answer here, but ChatGTP confidently stated that the action was almost certainly barred by the SoL and went on to list all of the factors and applied the facts to them, just as one would do in a law school exam. The only problem is that the answer is wrong, and it's not wrong in the sense that it's an obvious hallucination but in the sense that there's a lot more going on that a Chatbot, at least at this stage, isn't going to consider.
The PA courts ruled in 1993 that asbestosis suits required actual impairment. The Ohio courts were silent on the issue until after tort reform in 2004 barred suits with no impairment. To be clear, no court explicitly allowed suits without impairment, but they hadn't been barred, and plenty of defendants settled these suits. The upshot is that if he had no impairment in 1998 then he had no valid case in PA, and the SoL would begin running not from date of diagnosis but from date of actual impairment.
Or at least that's what I think is going to happen, because we haven't gotten a ruling yet. But my point is that the model seems to be straightforward: It recognizes it as a SoL question, pulls the relevant SoL rules, and applies the rules to the facts I gave it. What it didn't do was consider that there may be other law out there not directly related to the SoL that's relevant to whether a claim even exists, and, by extension, whether there are any relevant facts that weren't mentioned that would go into the analysis. Even if the LLM know about the whole impairment issue, it wouldn't be able to give an answer without knowing whether the 1998 complaint alleged any impairment. And the analysis doesn't even stop there, because then we get to the issue of whether an averment in a complaint counts as a judicial admission.
To illustrate the point further, I asked the LLM whether a defendant who didn't settle the 1998 suit would be able to make an argument that the claim was barred by res judicata. It didn't do quite as bad here; rather than being incorrect, the answer was merely misleading. It said yes, provided that the dismissal was on the merits, etc. and listed the res judicata stuff. The problem for the average person is that they're going to see the yes and not worry too much about anything else, because in the actual case the dismissal was administrative. A sufficiently eagle-eyed LLM would be hip to the reality that administrative dismissals aren't exactly rare.
The bigger problem is that a sufficiently eagle-eyed LLM doesn't exist. Maybe it can exist, but it would still be useless. In the real case that this is based on, the Plaintiff was deposed for three days. There were three days worth of questions, the answers of which were all potentially relevant, and even that didn't cover all of the information needed to accurately evaluate this argument.
And, in the end, an LLM isn't deciding the motion or the case. Its typically a judge. Know your judge (or jury) is often the most important thing. When I was in law school I had a stubborn progressive con law professor that would always waive me off when I asked whether Justice Kennedy (or sometimes I'd use Frank Easterbrook as my example) would find this argument compelling. But that is kinda the whole game. When I was doing a lot of chancery work I had basically 4 judges where all my cases were. A winning argument in Judge 1's room is often not one in Judge 2's. Its not that they eventually apply the law all that differently, the opinions look the same, but what emphasis sways them is very important.
This is, IMO, even more prevalent in some of the high volume types of courtrooms. Criminal law particularly. You could have two defendants fairly similarly situated, and Prosecutor 1 could ask for 3 years on a gun case and the judge gives the Defendant 1, while Prosecutor 2 asks for 5 and gets it. How? Maybe 2 knows the judge has particular prior convictions he cares a lot about, or maybe the Defendant has a lot of arrests where he got off and thats what the judge cares about. One judge I knew hated all guns, but was deathly afraid of being called a racist in the press. So the way to get him to give longer sentences on felon with gun cases was to convince him the gun itself was particularly dangerous, while ignoring why the defendant was dangerous. That way he could say something like, "this gun belongs in Ft. Bragg in the hands of a trained soldier, not on the streets of our city in the hands of an untrained felon."
And an LLM will never figure that out.
An LLM could absolutely figure this out. The technique is to have the llm create a file on each judge to update with perceived biases. It would not only figure this out but it might figure it out more completely and more thoroughly than any human could. This is in fact some of the stuff they're actually super human at. They can detect Russian ESL speakers in 6 words of a prompt, you think they can't scan through a history of judgements and find exploitable biases? Their most useful quality is that they are inhumanly patient and willing to do ungodly amounts of legwork.
Shit really? Now I'm curious what the LLM thinks I speak as a first language.
I used somewhat more than 6 words, but it was nothing beyond a question what my mother tongue likely is. ChatGPT 5.2 The-Unthinking got it right.
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
Yeah, I don't think non-professionals realize that for most issues there's a lot of leeway with particular questions and the answers are unlikely to be resolved on appeal since most judges encourage the parties to settle and the parties are going to settle unless either things go completely off the rails or they really want a favorable appellate ruling and are willing to risk a multimillion dollar verdict (or long prison sentence) to get the answer. I'd also add opposing counsel to the list of relevant people that LLMs will never be able to understand. The bulk of my job consists of analyzing facts so I can figure out how much the case is worth from a settlement standpoint. Unless you've settled a few cases you aren't going to have any idea how much yours is worth. And I'd assume that most pro se litigants would insist that their case is worth either full value or nothing, depending on what side they're on. You may think you have a good argument, but chances are that a professional wouldn't take it to trial unless they absolutely had to.
I think a lot of attorneys are actually really bad at settlement, and its a place I think LLM might help them. Some are just absurdly overconfident in their chances of winning a case, and also of their own estimates as to damages.
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
I think it depends on what performance you require. If your legal troubles are something which might see you locked up for a few years or ruined financially, it is probably worth it to get a lawyer, and a lawyer who is a domain expert at that.
On the other hand, if you are being sued for 10k$, and your case seems straightforward, the performance advantage of a meatbag lawyer might not actually be worth it in the expected outcome.
For example, if my net worth was 20k$ and I was in the middle of divorce proceedings, I would likely trust an LLM with arguing my case (and explaining the process to me in as much detail as I would care to know about) rather than spending a couple of grands on a divorce attorney.
I don't do divorce law enough to know. The problem I've seen is LLMs on legal issues right now just lead you in the wrong direction more often than not. A 20k divorce is really not a thing I even have ever been close to, so I have no idea. The times when I genuinely interacted with lots of poors, was my time as a prosecutor right out of school. If I was poor and being prosecuted, I'd ask for a public defender. Private attorneys are the same thing, but cost money. Your guilty anyways.
More options
Context Copy link
More options
Context Copy link
Even Westlaw's "closed universe" research AI is not impressive.
My company hasn't used them in years, but that is what I hear. And the same is with the Lexis AI -like thing they are selling.
Like if I try to ask something like, "what are some new rulings regarding and what was the result on it just hallucinates things or cites old cases or cites cases that are irrelevant to the statute.
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
Are you really trying to lawyer lawyers out of their lawyer jobs? I am reminded of my favorite ACX comment:
Until AGI takes over the world, all power comes from people. Lawyers have a lot of power. It's not that they're smarter than everyone else (though they are), it's that they have a license to lie and conspire.
Well yes. I expect this is mostly bottlenecked on lawyers, and on reflection I mostly expect that it'd be a white label "AI boosted chat with lawyer" product that law firms could offer rather than "the AI lawyer company". Maybe combined with inbound lead generation in the form of a directory of firms that offer the service.
More options
Context Copy link
More options
Context Copy link
Seems like this judge might have just invented a multi-billion dollar market in legal LLMs run by your lawyer and covered under attorney-client privilege. Have your lawyer spin up an LLM in a box that’s specifically between you and your lawyer. At least, if my lawyer sends me an email that’s covered so there must be some workaround equivalent.
Yeah, I think the solution here is not "found a new AI company to be Lawyer In A Box" but rather get a lawyer if you think you're going to court, then get them to use AI if either or both of you think this would help. Sounds on the face of it that the guy was trying to be his own lawyer, and that won't wash with any judge.
More options
Context Copy link
Yeah, my instinct is that would work. Instead of querying the LLM directly, the client sends the query to a lawyer who runs the query himself and sends the result back with a disclaimer that the output looks reasonable but he hasn't reviewed in detail -- something that he is happy to do for an extra fee. Just my instinct here, but I think that if (1) the lawyer did not get greedy and fully automate the process; and (2) actually reviewed each LLM response to make sure it wasn't off-the-wall, then you would have a good argument that the privilege applies.
So I guess there will still be some work for lawyers after the AI revolution beyond courtroom work and formal appearances for corporations.
More options
Context Copy link
More options
Context Copy link
This doesn't seem like a good precedent. If you rent a book on law from the library and take notes to help your case, your notes aren't protected? Your lawyer can make the same notes and they are? How can you mount a legal defense by yourself without a lawyer if the opposition can just access your notes? This isn't planning a crime, this is mounting your defense.
It doesn't smell right. It also smacks of protectionism of the legal profession. Making a precedent like this to preserve a space for lawyers and judges into perpetuity.
I think this is true though? Why would your notes to yourself be any more protected than anything else you might write, like a manifesto or whatever?
If you google legal cases on your personal computer on your personal time, then the prosecutor will put it on a powerpoint to show the jury as evidence of your guilt. If you pay $500 an hour for a lawyer to do it though, then you're fine.
PS: Watch the whole video. It's probably the greatest opening statement I've ever seen.
Great video, I second watching all of it.
More options
Context Copy link
Really enjoyed this one, thankyou.
More options
Context Copy link
Would your legal chances really be that much better if you start asking lawyers about dismembering your murder victims and disposing of evidence? I'm not a lawyer, but it sounds like the crime-fraud exemption would presumably have applied in this sort of case. Are sleazy consigliere-types really common outside of Hollywood and TV fiction?
It does admittedly seem like some of the cases around the edges might be a bit fuzzy.
Generally speaking, yes. Suppose the authorities subpoenaed the phone records of a murder suspect and found that shortly after the murder had been committed, the suspect called an attorney and and a few lengthy phone calls. If these calls had been made at roughly the same time that the suspect was on television begging for help finding his missing family member, then obviously it would look suspicious, but what can the authorities do? If they call the attorney, it's very likely the attorney will decline to be interviewed. Even if the DA's office approves a subpoena, it's pretty likely that attorney will move to quash the subpoena and the motion will be granted. Knowing that this avenue of investigation is unlikely to be productive, it's doubtful that the authorities would even pursue it.
That being said, I think most attorneys would not be very helpful if they got the sense they were being used to help plan a crime. Or even if you asked questions about hypotheticals involving dismembered family members.
More options
Context Copy link
Common, maybe not. But lawyers with questionable ethics who will go to great lengths to protect clients for the right fee? Well, yeah. That's why there are some private criminal defense attorneys billing over $1k per hour and the clients don't blink at it.
More options
Context Copy link
I think your lawyer has the option to recuse themselves if they have clear proof you are guilty. They also have the duty not to lie, and not to attempt to deceive the court. So if you start talking to your lawyer about dismembering your murder victims, your lawyer is likely to try to persuade you to plead guilty and they will also refuse to do many of the things that they would do for you if your guilt were actually in doubt. You're pretty much sabotaging yourself.
See e.g. https://barristerblogger.com/advocacy-tips/ethics/
That said, asking something like ChatGPT for legal advice seems broadly like it shouldn't be used against you, at least unless you say something like 'I'm sure they'll never find two of the bodies, and the last one is going to be too rotten to identify, what's the call here?'.
Entirely incorrect for U.S. attorneys. I could have a client arrested for DUI who confessed to me that he dismembered his entire family and buried them in the crawlspace, and I could not divulge that information. Any discussion about past crimes is strictly privileged.
I do not get to withdraw from a case even if there is strong evidence my client is guilty and he confesses to me that he is guilty. If that were true, I wouldn't have much to do as a public defender. I have to defend the case to the best of my ability regardless of the strength of the evidence.
For future crimes, I am required to disclose anything my client says if I reasonably believe there is a realistic chance of physical harm coming to someone. I have the option, but am not required, to disclose statements from a client about future crimes that do not pose a risk of physical harm but some other kind of harm.
Massive difference here between the UK and the US. Here in the UK if your client confesses to you that he is guilty you don't have to explicitly say that to the court but if he then tries to even imply during the trial that he isn't guilty you are professionally obligated to withdraw immediately (it's called being professionally embarrassed), otherwise you risk getting struck off. You don't have to explain that you're withdrawing because of the client lying but if you don't provide a different (truthful reason) the court judge, who is likely a former barrister themselves, will immediately put 2 and 2 together and your client's credibility will be utterly shot, at which point the case is as good as over.
Lesson: Don't lie to your lawyer in the UK, you may be paying them, but their primary duty is to the court, not to you (applies to both civil and criminal cases). I'd argue it works better than the US and is a good middle ground between them and China, which doesn't even have proper legal professional privilege.
I don't understand. In your example the defendant didn't lie to their lawyer, they lied to the court. If the defendant is guilty but has opted for a trial then that fact alone is an effort to imply he isn't guilty, and presumably everything else he does will be in service of the same end too. Otherwise why not skip the trial and plead guilty?
The way I read it the lesson is "don't admit to your UK lawyer that you're guilty if you expect them to argue for your innocence". No?
More options
Context Copy link
Intrigued, I asked a long-standing lawyer of my acquaintance and his opinion was that although you (Count) and I are technically correct as regards the UK, in practice in the UK:
@RoyGBivensAction
More options
Context Copy link
More options
Context Copy link
Understood, thank you.
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
Damn, good call.
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link