Angry judge roasts Biglaw lawyers for their "collective debacle" filing a brief where "~9 of the 27 legal citations in the 10-page brief were incorrect in some way" due to Generative AI
https://digitalcommons.law.scu.edu/cgi/viewcontent.cgi?article=3920&context=historical
Lawyers will pay $31k for their sloppiness 🤖😵
https://digitalcommons.law.scu.edu/cgi/viewcontent.cgi?article=3920&context=historical
Lawyers will pay $31k for their sloppiness 🤖😵
Comments
I've been out of practice for a while, but Lexis and WestLaw make automated, non-LLM citation checkers right? Do they not work?
https://www.abajournal.com/web/article/judge-imposes-sanctions-for-collective-debacle-involving-ai-hallucinations-and-2-law-firms-including-k
These lawyers screwed up repeatedly and intentionally, why give them a slap on the wrist?
That would be a sufficient deterrent. After the first ambulance chaser gets disbarred, the rest will think twice before using AI.
I am in the "AI is coming, let's deal with it" camp, so my attitude is "make the tool better" instead of just rolling my eyes at people using it.
That being said,
True "reasoning" and error-checking will have to come from a different, not-yet-invented tech.
https://techblog.comsoc.org/2025/05/10/nyt-ai-is-getting-smarter-but-hallucinations-are-getting-worse/
The flaw is the whole "prompt" thing - they still need to be told WHAT to say. An LLM is the ability to speak, NOT to think.
A true Artificial Intelligence might use an LLM as its input/output interface, but the logic/reasoning model is still hypothetical - and will never "evolve" from LLM tech.
What the AI generated looks so close to actual legal arguments that the lawyers submitting it as such were fooled! Considering that the literal purpose and goal of AI is to make things that look exactly as expected, it feels 100% accurate!
Now, the lawyers need to be sued out of existence and disbarred for submitting those AS real legal documents, but that's a different matter