March 20, 2026

Your prompts, now courtroom drama

CAIveat Emptor: What You Tell AI Can and Will Be Used Against You

Judge says your AI chats aren’t private — commenters split between “duh” and “doom”

TLDR: A federal judge said chats with AI like Claude aren’t protected like talks with a lawyer, meaning your prompts could be used in court. Commenters split: some say stop pasting secrets, others fear creepy tracking apps more, and a few ask if local, offline AI could dodge the risk—this matters for anyone using AI at work.

The internet is clutching pearls and cracking jokes after a New York judge ruled that telling your secrets to an AI like Claude isn’t protected like telling a lawyer. Translation: those late-night “uh-oh, how do I not go to jail?” prompts can show up in court. Cue panic, eye-rolls, and memes.

On one side, the pragmatists are screaming: treat AI like a stranger, not a diary. As one pro put it, it’s a third-party service—so don’t paste secrets in your prompts. Others are like, “Wrong villain!” Havoc says the real creep factor is apps reading your soul by how long you hover on a pic. Meanwhile, keybored paints the bleak modern vibe: either get exploited, become a tech wizard, or go off-grid and get nagged for it. The existential dread is real.

Then there’s the meta-drama: someone shouted “7 ads above the fold,” roasting the ad-stuffed site carrying the news while we argue privacy. And the nerdier corner is asking: what if you use a local AI on your own device? What if there’s an AI tool built just for lawyers? The ruling here doesn’t answer that—yet. But one thing’s clear: bosses say “Use AI!” The court says “Careful what you say.” And the comments say “Welcome to the surveillance sitcom.”

Key Points

  • On Feb. 10, 2026, Judge Jed S. Rakoff (SDNY) ruled that a defendant’s consumer AI queries were not protected by attorney-client privilege or the work product doctrine.
  • The ruling arose from United States v. Heppner, where the FBI found documents of the defendant’s communications with the AI platform Claude.
  • The defendant used Claude to explore legal defense strategies after learning he was under investigation and without counsel’s prompting.
  • The court held privilege did not apply because Claude is not an attorney, a “trusting human relationship” cannot exist with an AI platform, and communications were not confidential due to platform data policies.
  • The article urges businesses to rapidly implement AI acceptable use policies to reduce litigation risk and protect trade secrets and sensitive information.

Hottest takes

insta and TikTok algos that deduce things from you hovering over a pic a imperceptible fraction of a second longer scare me more — Havoc
Man 7 ads above the fold — dwedge
Either be a regular, naive user who wants to use tech as a means to an end and get exploited at every turn — keybored
Made with <3 by @siedrix and @shesho from CDMX. Powered by Forge&Hive.