Can You Sue an AI for Defamation? Where the Law Stands

One defamation claim over AI-generated output has reached judgment in the United States, and the AI company won. In Walters v. OpenAI the Georgia court granted summary judgment to OpenAI in May 2025, holding that a reasonable reader in that context would not have taken the output as a statement of fact and that the plaintiff had not met the required fault standard. Other cases are still moving. As of now, litigation is not a reliable route to a corrected answer.

This is a plain-language explainer rather than legal advice. If a false statement about you is causing real harm, speak to a lawyer in your jurisdiction. What follows is what the record currently shows, and why the practical work sits elsewhere.

The case everybody cites

Mark Walters hosts a radio show in Georgia. A journalist asked ChatGPT to summarize a lawsuit, and the output described Walters as a defendant accused of embezzling funds from a nonprofit. He was not a party to that case and had not been accused of anything. The journalist did not publish it, and he contacted Walters instead.

Walters sued OpenAI for defamation. The case became the first of its kind to reach a substantive ruling in an American court.

In May 2025 the Superior Court of Gwinnett County granted OpenAI summary judgment. Three threads ran through the reasoning.

• Context defeated the claim of fact. The court accepted that a reasonable reader, seeing the disclaimers and understanding the circumstances of the exchange, would not have treated the output as an assertion of fact about Walters.
• The recipient did not believe it. The journalist checked and did not publish, which weakened the argument that the statement was understood as true.
• Fault was not established. Defamation requires a fault standard, and the court found the plaintiff had not met it against a company that discloses the possibility of inaccurate output and works to reduce it.

What the ruling does and does not settle

It is one trial-level decision in one state, on one set of facts. It does not hold that AI output can never be defamatory. A different case with wider publication, a recipient who believed the statement, and evidence that the company knew about a specific recurring error could come out differently.

What it does establish is that the disclaimer-plus-context argument works, at least once, and that plaintiffs face a harder path than they would against a publisher. Anyone selling you a legal strategy as the answer to an AI reputation problem should be asked what they think happens after this judgment.

Why suing rarely gets you what you want anyway

Set aside the odds for a moment. Consider what a win would actually deliver.

Damages compensate. They do not edit a model. There is no mechanism by which a court order removes a claim from training weights, and no engine offers a per-company correction interface for a judge to point at. A successful claim might produce money and a public record, and the answer your buyer sees next week could still be the same answer.

Litigation is also slow and loud. In most of the categories where AI answers cost the most, and financial services and healthcare sit at the top of that list, a public dispute about whether you are trustworthy produces exactly the material that engines retrieve later. You can win the case and enrich the source pool that generated the problem.

What is still moving

Other claims involving AI-generated statements about named individuals have been filed in the United States and in Europe, and regulators in some jurisdictions have opened questions about accuracy obligations under data protection law, which is a different lever from defamation. Data protection frameworks that include a right to rectification of inaccurate personal data are a live area, and how that applies to a generated answer rather than a stored record has not been resolved.

The position may change. It has not changed yet.

What to do instead, or alongside

The work that changes answers is supply-side, and it is available now regardless of what the courts decide later.

• Preserve the evidence. If you may ever litigate, or complain to a regulator, you need dated records of the exact prompt, the engine, and the full answer text. Nothing is retained on the engine side, so an answer you did not save is gone. Our clients get every run stored in the client portal with its date for this reason.
• Find out whether it is retrieval or memory. Ask the same question with search on and off. If the claim survives without search, no page you publish removes it today. If it only appears with search, a specific source is producing it and the fix is concrete. The full test is in our piece on what to do when ChatGPT is wrong about your company.
• Correct the sources. Roughly 86 percent of AI citations in one analysis of 6.8 million traced back to first-party pages and listings. If the true version of the fact is not published clearly and dated, the engine has nothing better to reach for.
• Repair the entity record. A large share of genuinely false statements about individuals come from confusion with a namesake rather than from invention. That is fixed with structured data and authoritative records, not with a lawyer.

Frequently asked questions

Has anyone won a defamation case against an AI company?

No claim of this type has produced a plaintiff verdict on the merits in the United States as of now. Walters v. OpenAI, the first to reach judgment, went to OpenAI on summary judgment in May 2025.

Does the disclaimer protect them completely?

It was part of the reasoning in one case rather than a blanket shield. Context, how widely the statement spread, whether the recipient believed it, and what the company knew all mattered.

Can I use data protection law instead?

In jurisdictions with a right to rectification of inaccurate personal data, this is an open question and worth asking a lawyer about. It applies to individuals rather than to companies, and how it operates against a generated answer rather than a stored record has not been settled.

Should I send a legal letter to the AI company?

That is a question for your lawyer. What we would say is that it should not be the only thing you do, because even a cooperative response does not come with a mechanism to correct a specific answer.

The practical position

Treat the legal route as something to keep open and not something to rely on. Preserve evidence properly, in case the position changes. Then spend the effort on the sources, because that is the part that moves.

Our AI reputation management service is built around that order of operations, with dated answer archives across six engines from the first month.

Related guides

• AI Search Is Changing Where Business Inquiries Come From: The Data for the Gulf and Egypt
• Google’s Search Generative Experience (SGE) and AI Overviews: What Egyptian Brands Need to Know in 2026
• GEO: Generative Engine Optimization – Your Guide to Ranking on ChatGPT
• Generative Engine Optimization (GEO) Services in Egypt: The 2026 Playbook