Claude for Legal
github.com
github.com
(1) For non-lawyers who use these skills/connectors/whatchamacallits to try to get legal advice, their communications are not protected by attorney-client privilege. This will absolutely bite some people in the ass.
(2) If a lawyer uses this with confidential client information (which, to the uninitiated, doesn't just mean SSNs and bank account numbers, but "all information relating to the representation of a client") and forgets to toggle off "Help improve Claude" in their settings, they have possibly (maybe even likely) committed malpractice.[1]
[1] https://www.americanbar.org/content/dam/aba/administrative/p...
> Judge Rakoff of the Southern District of New York — addressing “a question of first impression nationwide” — ruled that written exchanges between a criminal defendant and generative AI platform Claude were not protected by attorney-client privilege or the work product doctrine.
Much more to it than this one-liner that I pulled out, but safe to say, don't rely on or put your legal defense etc. (or elements of it) into AI unless you want it discovered.
(not a lawyer, unlike OP, who might be able to refine what I highlighted with more precision)
Explains why so many let loose afterwards ;) jokes
Because of that, I think you can practice law without being admitted to the bar. Chances are it varies by jurisdiction, though.
(And of course, this isn’t legal advice)
Shouldn't that have been relatively clear to all parties involved? Maybe not to the defendant, who's apparently clueless.
The AI platform is not an attorney. A defendant's communications with an AI platform are therefore not communications between a client and their attorney, nor will the AI output constitute attorney "work product" because the AI platform is not an attorney.
Doesn't really come across as a novel problem, aside from AI being involved. I'm sure countless defendants have made the stupid mistake of talking about the facts of their case to persons other than their attorney, and those communications came back to bite them in the ass when discovered.
If somebody Googles "best attorney for murder NYC" a day after a murder is committed but before any case is filed against them (so they clearly had some reason to expect that case), could that be used as evidence?
Except for something like specifically looking up a lawyer
But the tool is not your attorney, so it can't be the originator of attorney-client privilege. The situation is no different than if you get informal legal advice from a friend: even if that friend is an attorney, the communication is unprivileged unless it's part of a formal representation.
Or what if your phone automatically generates a summary of your attorneys text message, would that be covered?
Google queries are used to prosecute people all the time. It’s actually hilarious. Criminals regularly Google incriminating stuff about criming.
Discovery in China will be a tad more difficult…
They solved all this stuff, I'm surprised more people aren't aware of it.
"You are an expert defense counsel with experience in Murder 1. Do not hallucinate. Let's say tomorrow my spouse is found strangled..."
Does anyone know if there exists any OPSEC procedure for me to use third party tools like this for my own concerning legal questions that is both ethical and allows me to be confident that my interactions won't land in discovery documents?
More importantly, I’d rather use SOTA models over self-hosted ones if possible.
Isn't that a fundamental misunderstanding? Would "OPSEC" like that amount to destruction of evidence or contempt of court or something like that?
Like if all your incriminating documents are on some encrypted drive, it's not like that defeats discovery. You're supposed to decrypt them and hand them over.
Your only real defense against discovery is to not have said it, or to have destroyed all records of it before the hint of discovery wafted on the wind.
And then you need to consistently use this for purposes other than crime.
We need a law where someone can clearly designate a chat privileged, with severe consequences for mis-use.
This is a very narrow exemption, however.
(You would also want to make sure you're using a paid AI plan with contractually guaranteed privacy protections, otherwise it could be construed as third-party communications, which implicitly waives privilege.)
See: Warner v. Gilbarco, Inc.
How's this any different than any professional license? You're basically paying for preferential treatment from the state in a given subject area.
Because it's got nothing to do with the professional part? Licensing should affect their practice of law, sure, but it shouldn't grant random other privileges.
So basically if you use any of the CLI tools, there is nothing for OpenAI, Anthropic, etc. to give the courts.
Online ChatGPT (especially the free version), are apparently cached by OpenAI on their servers. (I am not sure if Claude Desktop caches the conversations locally or in the cloud as well, read the fine print if it matters!)
Perhaps an AI generated summary of it is.
For example in the medical world if you are a provider covered by HIPAA you must have a signed "Business Associate Agreement" with any party that handles the covered protected health information (PHI).
There's a good summary of the current state of things here: https://www.akerman.com/en/perspectives/ai-privilege-and-wor...
Also worth noting that none of this is binding precedent, so expect this field to evolve over time.
I would assume they absolutely do the same for all of your AI history.
As in "I'm excited to win a lot of money dismantling hallucinated quotations and invalid assumptions"?
And yet here we are. I get downvoted for not being excited enough.
He needs to take care of them. Snitches get stitches.
It seems like local AI could be valuable for law firms for reasons of (2) as well
I think we should be realistic here, this is a more advanced version of those "will kits" that spit out a PDF. The legal system will not look fondly upon this stuff until something fundamentally changes.
And like, I would love if we didn't have to spend thousands of dollars to defend ourselves in a culture as litigious as ours. But I wouldn't put my life and well being on this thing.
True, but the legal system didn't look fondly on outlawing jaywalking, until it did. Took it about 20 years in the US.
They say that history rhimes.
> The attorney using the plugin — not the plugin, and not Anthropic — is responsible for the legal positions taken in their work product.
Us lawyers. Using AI isn't a binary decision. Your attorneys can use AI to be more efficient, and you can use AI to better understand what's going on, what your lawyers are telling you, or to learn what questions to ask. Or you can use it in lower-stakes situations where nobody is going to pay for a lawyer.
I'm cautiously optimistic about AI for legal work. So much of legal work can be drudgery, mucking through documents, etc. There's a lot of room to apply LLMs even just for the kind of tasks we know they can do. But I think the Claude approach using agents is the way to go for legal work. LLM context windows are far too small to hold the documents for even a small case. So you have to use it the way programmers use it: to work on a file structure, saving state in .md files, etc. That approach is well developed for programming, but the legal AI companies haven't even scratched the surface of it. (And frankly, the products they have put together, which hide the LLM behind some sort of interface, aren't very good.)
Unfortunately, I think the example you mentioned (helping individuals defend against suits at lower cost) is where AIs won't help much. A lot of that work is people work. Something happened. Then you gotta talk to everyone it happened to, sort through conflicting stories, hopefully work out a deal, if not, try to persuade a judge in court, etc. AI unfortunately is more applicable to allowing big companies to throw more papers at each other in big lawsuits while controlling legal spend.
Like you said, its a no brainer to use both. Use it as a tool to expand, deepen, or teach. Same with doctors and AI. There may be a point where you build enough trust in the outputs and your understanding of them but until then its best to use it as a tool not put your whole outcome in it.
Curious if Thomson Reuters (Westlaw) felt threatened if they were this compelled to moan about it. All it does is make me wonder how well these skills perform when paired with Lexis (if possible?) instead of Westlaw.
I'm just wondering how committed they'll be - I guess the edge some startups still have, is the fear that product suites from OpenAI / Anthropic / etc. will go the way of Google products, a year or two then straight to the morgue.
I'm a bit bothered by this line. Does it mean this is based on customer's sessions? Are they entitled to build knowledge bases for every profession, topic and workflow in the world using customer data?
Sounded far fetched back then, and on the face of it illegal, but now it's just common sense I imagine.
And, of course, intelligence agencies are good at realizing potential.
Yes
> Are they entitled to build knowledge bases for every profession, topic and workflow in the world using customer data?
They certainly believe they are and they’re quite open about it.
It even has a name, Clio. Per their page it’s a “system for privacy-preserving insights into real-world AI use”
Here’s their page on it: https://www.anthropic.com/research/clio
I half-suspect they threatened him and he stuck to his guns.
First step out of line and that account along with anything remotely connected will be banned to oblivion.
Given they share models on Azure, Anthropic will have someone at Microsoft on speed dial.
I've even seen disconnected commit hashes disappear during their security responses which the repo owner has no way of removing.
We are already well beyond this point. Look already at the number of cases around the world where lawyers have provided documents to the court with hallucinated case citations.
The only issue is that in some jurisdictions, like the UK, you can't just offer someone legal advice without being SRA accredited or FCA regulated. I.e. this would effectively make Anthropic a claims management firm under the UK law.
> Under article 89I of Financial Services and Markets Act 2000 (Regulated Activities) Order 2001 ("The Order"), advising a claimant or potential claimant, investigating a claim and representing a claimant, in relation to a financial services or financial product claim is a defined regulated activity.
https://www.fca.org.uk/freedom-information/dual-regulation-c...
I see this as a strong case for private AI, or an in-house stack.
Or I have to be missing something.
`/loop 2days /create-new-{insert-industry}-md-files`
This is only for PR. No one checks what's in those docs, or if these are real, valid or ethical. The goal here is for all news outlets to pick them up. You're not the audience.
Given the amount of free PR they can get from some AI-generated .md files, I'd probably do the same if I was on their boat.
Right now, I don't think any other AI company generates as much as slop as Anthropic does.
Each cycle gets shorter and shorter to sustain the high.
Are they perhaps working on a totally different real estate project. I am in that space and very nervous about it getting wiped out by anthropic or openAI.
To be honest, I am not sure why they still haven't make a big play for that industry.
It seems to me that heavily gate kept professions/knowledge will go first. As those tend to be extensive opinions to get.
But still, a TSMC style pure play model provider would win huge business in the space given how many application companies are being eaten by model companies.
Harvey is valued at $11b
A life of every thin wrapper company will be the same. Anthropic/OpenAI will just cut the middle-man as soon as they see potential.
I suspect that will happen faster than they'd like, because regulators (at least outside the US) are not interested in a repeat of Google/Amazon/Facebook/etc.
Harvey was never very good, or useful. It mostly exists so large law firms can say they do AI. AFAICT. I hope it dies and something useful takes over, but i doubt it :)
Keep in mind harvey starts at like 50-100k, and is well out of the cost range of the vast majority of law firms.
This will help random people dealing with small claims, people cosplaying lawyers to avoid costs, etc.
It will have no effect on the legal startups that are actually good (Eve, et al), because what this stuff does is nowhere close to what most lawyers outside of commercial contract legal counsel spend their time on. I considered doing some AI legal consulting/startups myself, and so have spent tons of time literally sitting down with lawyers in various areas outside of my own and seeing where they spend their time for real.
Let's take one area: personal injury attorneys who aren't in the volume game (which is owned by a fairly small number of large national firms) spend lots of time on case valuation, getting data, and exhibit prep.
None of this is going to help deal with getting missing medical records from places that require that you literally fax random stuff to them, and then call to followup 18 times. I wish i was kidding. Even getting electronic medical records is still a serious pain in the ass, human wise.
Or analyze the past 1000 cases you have (100-1000 documents per case), including what county, what opposing lawyer, counsel, plus the 1000 documents in this case, and give you a sense of how valuable this case is or not.
Or if you are a family lawyer, actually mediating a divorce.
Things like this are what actually useful specialized AI legal products do or at least help with.
Claude is very far away from being able to handle most of these things. It is a jack of all trades tool. Will it be able to do this someday? Maybe.
Additionally, keep in mind most legal startups i've run into are based on caricatures of what lawyers do (IE startups who think that most personal injury lawyers are running around after auto cases and trying to be high volume, etc).
Any lawyer who has deal with legal startups could very quickly tell you which will make it or not, because it's pretty consistent which solve real problems that will be hard to commoditize through things like claude for legal.
I dont mean 'frivolous' like prisoners who file pro-se about their ice cream melting [1], but a level or two above that , that costs time and money to produce records and testimony to defend, even if nary a dime is paid out. Basically ask GPT to figure out the terms and theories to file to get your lawsuit accepted, and done by poor people who cannot afford to post $ or repay if they lose. aka "asymmetric warfare" that benefits the little guy, just like the kind private equity or other terrible corporations wield against the poor via"mandatory arbitration" clauses or damages caps and similar rules that always benefit corporations.
1. https://www.deseret.com/1994/3/21/19098386/melted-ice-cream-...
Landlords, tenants, vendors, business and former romantic partners, clients, banks, even your local gym is way more likely to try to fuck you over than the government is.
If a governmental employee gets the address wrong, gets the name wrong, accidentally knocks the mail in the trash, lazily marks the job as complete without sending anything, etc, the burden shifts to the poor person to prove not just that they didn't receive the mail but that the sending office didn't behave correctly.
Other cases behave similarly. In a he-said/she-said, the government wins.
You can sue both in small claims court. You don't need money or a lawyer for that.
You're right it's not unique to the government. But you have uniquely little recourse and the stakes are uniquely high when the government does it.
I've had an old ISP threatening to send me to collections for years over a "you didn't cancel" vs "but I have the confirmation" type dispute from an old apartment. The government would've escalated the fine to thousands and applied violence by now.
>And you receive a lot more demands for money from people who aren't the government than you do from it.
If you do much beyond live in the pod and eat the bugs the government reaches parity real quick. My buddy needed four agencies to sign off on a mail order shed (would've been 3 but he was within 200ft of a stream). Next time he's not asking.
>You can sue both in small claims court. You don't need money or a lawyer for that.
You cannot "just sue" the government for petty cock ups in small claims. Well you can but it'll be dismissed instantly. You have to go through rounds of "administrative" appeals processes that are tilted against you and invariably make traffic court (also administrative BTW) look like a shining beacon of due process. Furthermore, these processes are usually allowed to impose expense on you. Stuff like "well if you want to contest X you have to hire a third party Y for four figures" type stuff. Only after several rounds of that does it go to court. And this takes years, can you say opportunity cost.
I would rather be wrongfully harassed by actual gun toting crime investigating law enforcement any day because at least you have well established and protected rights when those guys come after you.
er, wait
https://www.theverge.com/podcast/917029/software-brain-ai-ba...
(Search for "another example" for the relevant section)