That ChatGPT Legal Advice You Asked For? It Can Be Used Against You in Court

You’ve probably done it. Pasted a contract into ChatGPT and asked “does this look okay?” Or described a workplace situation to Claude and asked whether you have a case. Millions of people treat AI chatbots as informal legal advisors every day. A federal court just ruled that those conversations are fair game as evidence. In US v. Heppner, the court held that attorney-client privilege does not extend to AI — and the implications are enormous.

Attorney-Client Privilege Has a Hard Boundary

Attorney-client privilege is one of the most powerful protections in the American legal system. What you tell your lawyer stays between you and your lawyer — no court can force its disclosure. It’s the reason people can be fully honest with their counsel without fear.

But the privilege has always had strict requirements. The conversation must be with a licensed attorney, conducted for the purpose of obtaining legal advice, and kept confidential. In US v. Heppner, the court applied these criteria and reached a straightforward conclusion: AI chatbots are not attorneys. No state bar has ever licensed one. So no matter how legal the question feels, a conversation with an AI carries no more privilege than venting to a friend over drinks.

And there’s a second problem. Most AI services store your conversations on their servers. Even the confidentiality requirement starts to crumble.

The Logic Is Simple — and That’s What Makes It Dangerous

The ruling’s reasoning isn’t novel or creative. That’s precisely what makes it so hard to challenge. Courts didn’t invent a new doctrine or make a controversial leap. They applied existing privilege law to a new technology and found it didn’t qualify. The three-part test for privilege — licensed attorney, legal purpose, confidentiality — has been settled law for decades. AI fails at step one.

Legally, asking ChatGPT for advice is equivalent to asking your neighbor who “watches a lot of legal dramas.” Both conversations can be subpoenaed, introduced as exhibits, and used to establish what you knew and when you knew it.

The Real-World Scenarios Are Alarming

The blast radius of this ruling extends far beyond criminal defense.

Criminal cases. If a defendant asked an AI “what’s the penalty for this kind of thing?” before or after an alleged crime, prosecutors can request those chat logs. That query alone could be used to establish awareness or intent — devastating in court.

Corporate liability. Imagine an employee asking an AI about the company’s environmental compliance risks or potential labor law violations. In litigation, those conversations become discoverable. They could serve as evidence that the company knew about problems and failed to act.

Personal disputes. Divorce proceedings, landlord-tenant fights, employment disputes — in any civil case, opposing counsel can seek your AI chat history. The questions you asked become a roadmap of your legal thinking and vulnerabilities.

The Legal Community Is Split

Not everyone agrees on what to do about this. Some legal scholars say the ruling is obvious and correct — privilege exists to protect the attorney-client relationship, and there is no such relationship with a chatbot. Case closed.

Others argue the law needs to catch up with reality. AI has become a de facto legal resource for millions of people who can’t afford attorneys. Self-represented litigants — people navigating the court system without a lawyer — are hit hardest. The very people who rely on AI because they can’t access traditional legal counsel are now the most exposed. There’s a bitter irony in a system where those who can afford lawyers get protection, while those who can’t get their AI conversations used against them.

No new legislative framework has emerged yet. The gap between how people use AI and how the law treats it is wide and growing.

What You Should Actually Do

This doesn’t mean you should stop using AI for legal questions entirely. It means you should be deliberate about it.

For anything with real stakes — potential litigation, criminal exposure, sensitive business decisions — talk to an actual attorney. That’s the only way to get privilege protection. When you do use AI for legal questions, keep them general. Don’t include names, dates, specific incidents, or anything that identifies your situation. Treat every AI conversation as potentially public.

If you run a company, this ruling demands a clear internal policy. Employees should not be discussing the organization’s legal risks with AI tools. Full stop.

The law hasn’t kept pace with how deeply AI has embedded itself in daily decision-making. US v. Heppner is the first major marker of that gap. We live in an era where you can ask an AI anything — but “can” and “should” have never been further apart.