Lawyer.exe | Trial by AI

AI can be a great tool in the legal field, for lawyers and clients alike. But you need to be aware of its limitations and potential pitfalls. Clients: AI CANNOT replace your lawyer. Lawyers: AI CANNOT replace your brain. Use it right.

Just saw this in a @JamesClear newsletter: "Your response to a challenge defines you more than the challenge itself."

@fantony_francis Disciplinary committees filled with lawyers who couldn’t get real work 😂. Ain’t that the truth.

@litigationai So we'll need more disciplinary committees ( filled with lawyers who couldn't get real work) and AI-illiterate judges to impose sanctions. CA has that money too, just raise taxes. More wasted taxpayer dollars for a stochastic parrots who serve their masters with wrong advice.

Law schools are adding AI courses. The California Bar is proposing AI competence requirements in the ethics rules with disciplinary teeth. Courts are sanctioning attorneys who don't verify output. The profession is coming at this from three directions at once. It's a structural problem worth recognizing.

@AMandoSch @pop_tartlet Twexcellent.

@pop_tartlet Kids like Twotes

It's been a little busy and no chil and I've fallen a bit behind on the Twittplies and Twittilikes. For that I apologize. I will catch up but in the meantime will keep the Twotes coming.

ABA Formal Opinion 512 says AI literacy is competence under Rule 1.1. California is writing that into its ethics rules with disciplinary teeth. Florida just enacted a certification requirement effective June 15. The profession had three years of advisory opinions and sanctions warnings. Now the rules themselves are changing. The attorneys who treated this as a theoretical ethics question are about to find out it isn't.

If you're a litigator, it better be the latter. The courtroom doesn't care about your excuses.

When everything goes sideways, you either hide, deflect, and blame, or you take control of the situation. That choice matters more than what broke in the first place.

83% of attorneys now have access to AI tools. 22% trust them enough to rely on output without heavy verification. That 61-point gap is the legal profession's actual AI problem, and nobody selling a $50/month subscription is closing it. They sell access to the model. The workflow knowledge, the verification habits, the understanding of where AI fails in legal contexts—none of that ships with the product. The gap closes one of two ways: attorneys build competence through actual use and honest feedback loops, or they get sanctioned into it. The profession is currently doing both simultaneously.

@chiajy2000 Yeah that' what I get for scheduling a post when I should be asleep 🤣 Thanks man. Corrected!

@litigationai Oh, it wasn’t the user suing though

An insurance company sued OpenAI claiming ChatGPT gave a former claimant specific, actionable legal advice—and it was wrong—and they acted on it. The UPL question for AI products is not hypothetical anymore. The cases that define where "general information" ends and "legal advice" begins will shape what AI legal tools are permitted to do. First court to draw that line clearly draws it for everyone.

@chiajy2000 Nippon Life v. OpenAI case. law.stanford.edu/2026/03/07/des…

@litigationai Interesting - is there a link to source?

@CaseFunders I agree!

@litigationai Legal AI creates real value only when paired with verification, ethics, and accountability. As schools, bars, and courts move in the same direction, firms will need clear AI policies and disciplined review processes.

@gordon_cassie I literally thought the same thing as soon as I saw the link.

How many times have they announced an Agent now?

@DavidShulmanFL Yeah, that would definitely be bad. I think one of the better ways to counter that kind of nonsense is have AI clauses in your retainer agreements. The client is on full notice what you do and don’t do with AI. That serves to defang the anti-AI zealots.

@litigationai I’m very concerned that some of the rules could be written by reactionaries who want to assert that using AI for any client information is per se unethical

Oral argument prep: after the brief is filed, run the argument through Claude as the panel. Feed it both briefs and the relevant cases. Ask for the five hardest questions a skeptical judge would ask. Then answer each one out loud.

@TwistNH7 The lawyers "risk sanctions and problems like that" part. Not the "actively harming lawyers" part. The lawyers harmed by AI are doing it to themselves.

@TwistNH7 Yes, that is most certainly true.