By William D. Slicker
Recently, a prospective client provided me with a Smart Summary of a deposition by Veritext Legal Solutions in which the AI generated summary replaced the deponent’s name and with the name of the presiding judge. This resulted in the following section of the Summary:
The deposition discusses whether Judge [author has deleted the name] was offered a drug and alcohol test by the Department of Children and Families (DCF). Judge [Redacted] initially stated that they did not recall being offered a test but later mentioned a situation involving CPI [Name Also Redacted], who came to their residence and offered a drug test. Judge [Redacted] declined the test under advice of counsel but eventually agreed to take it after further discussion.
My good assistant, Amanda, told me that if Veritext Legal Solutions’ AI could make such a false statement, she wondered what other false information AI was capable of sending. So I looked into that issue.
There have been only a handful of defamation cases based on claims of AI defamation.
First came Battle v. Microsoft Corp.[1]. In that case, Jeffery Battle sued Microsoft because Microsoft CoPilot confused him with another individual with a similar name who was a convicted terrorist. Without addressing the merits, the court found that the plaintiff was bound by an arbitration clause, and referred the case to arbitration.
The second case was Walters v. Open AI, LLC.[2] In that case Mark Walters sued OpenAI, LLC because ChatGPT falsely claimed he had been accused of embezzling funds from the Second Amendment Foundation. Walters had a radio program advocating gun rights with an estimated 1.2 million listeners. He had authored books and articles on Second Amendment rights. He served on the Board of Directors of the Committee for the Right to Keep and Bear Arms. He was the east coast media spokesperson for the Second Amendment Foundation.
The court first held that the ChatGPT user knew that AI could produce hallucinations and had received disclaimers warning that mistaken output was a real possibility. Therefore, the user should have attempted to verify the information. The user admitted that within an hour and a half, he established that ChatGPT’s output was not true.
Next, the court found that there was no actual malice or even ordinary negligence. The evidence presented was that OpenAI used reasonable care in designing and releasing ChatGPT according to industry standards.
Thirdly, the court found that Walters had conceded that he had no actual damages.
With those three strikes, the court entered Summary Judgment in favor of OpenAI, LLC.[3]
In the third defamation case, Robert Starbuck learned that Meta’s Instagram and Facebook applications were falsely stating that Starbuck had pled guilty over disorderly conduct arising from the January 7, 2021 break-in at the Capital, and that he was advancing “Holocaust denialism”. Mr. Starbuck notified Meta about these false outputs. Yet nine months later, Meta continued publishing them. The case was settled.[4]
The fourth case involving false output was Jane Doe 1, Jane Done 2, and Jane Doe 3 v. X AI Corp and X AI, LLC. In that case, someone took the photos of three minor children that had been posted and used an AI program called GROK that allows someone to remove the clothes from a photo. The now nude photos with the original faces were then used for child pornography. The three minors sued Elon Musk’s X AI, Corp. and X AI, LLC on several theories including production and distribution of child pornography, design defect, and infliction of emotional distress.[5]
The fifth case is LTL LED, LLC d/b/a Wolf River Electric v. Google, LLC[6] which was initially removed from state court to federal court, but was remanded back to state court in January, 2026 and is still pending. The case is a defamation case that involves a Minnesota solar company that alleged it suffered between $110 million and $210 million in damages in reputational harm, diminished goodwill, and lost profits after Google’s AI Overview falsely stated that the Minnesota Attorney General had sued the company for deceptive sales practices, high pressure sales tactics, hidden fees, and installation issues.[7]
From the few cases, it appears that the courts may not apply human defamation law to AI. However, it may be possible to find redress through a design defect case or another theory.
[1] Battle v. Microsoft Corp., Case #23-cv-01822-LKG (D. MD. 2024)
[2] Walters v. OpenAI, LLC, Case #23-A-04860-2 (Sup. Ct. Gwinnett Cty, 2025)
[3] Starbuck v. Meta Platforms, Inc., Case #N25C-20-211-MAA (Del. Super. Ct., filed Oct. 2025)
[4] Id.
[5] Jane Doe 1, Jane Doe 2, and Jane Doe 3 v. X AI, Corp. and X AI, LLC, Case #5:26-cv-02246-SVK (N.D. Cal. 2026)
[6] LTL LED, LLC d/b/a Wolf River Electric v. Google, LLC., Case #0:25-cv-02394 (D. Minn.)
[7] LTL LED, LLC d/b/a Wolf River Electric vs. Google, LLC, Case #0:25-cv-02394 (D. Minn.) https://storage.courtlistener.com/recap/gov.uscourts.mnd.225722/gov.uscourts.mnd.225722.1.1.pdf
