A common, age-old legal question many business owners have is whether they can sue over a bad Yelp or Google review. But what if an artificial intelligence (AI) platform like ChatGPT, Gemini, or Claude slams your business? Can an AI chatbot be legally responsible if it generates false information about a real person or business?
As with most legal issues, the answer depends on the specifics of your situation and your state's laws.
Defamation laws protect people and businesses from damaging, untrue statements. Defamatory statements can take two forms: Libel if they’re written, and slander if they’re spoken. Although the exact rules vary by state, most defamation claims share the same core requirements, including proving that a false statement was shared with others and caused harm.
Whether the scathing AI output was based on real reviews or a “hallucination” can determine who or what may have legal liability. An AI hallucination occurs when a generative AI system outputs an answer that is factually incorrect or, in some cases, completely fabricated. Think: citations or quotes that are entirely made up.
Regardless of where the statement originated, a successful lawsuit still requires proving all the elements of defamation.
The 4 Elements of Defamation
To prove to a court that AI-generated content qualifies as defamatory, you still must show the traditional elements of defamation for non-public figures:
- The statement was false.
- The person or entity who made the statement knew it was false, or should have known it was false.
- The statement was shared with a third party, either through publication or oral communication.
- The statement caused reputational damage or harm.
If the statement falls within a category of defamation per se, you may not have to prove that it harmed you or your business. Courts presume such statements to be inherently damaging. This can include false statements involving sexual misconduct or committing a crime. However, not all states’ defamation laws recognize this type of defamation.
AI Said What About My Business?
Sometimes AI gets it way, way wrong. ChatGPT also raised some eyebrows when a Norwegian user provided his name as a prompt, and the tool drafted a compelling (yet wholly untrue) story about him being a convicted murderer.
While some AI hallucinations are good for a laugh, others can be devastating to individuals and businesses. But to sue an AI company for a hallucination, you must prove that the response meets the legal criteria for a “statement of fact.” At their core, AI tools work by recognizing patterns and predicting what is most likely to come next based on their training data. Because of this, it can be tricky to convince a court that a reasonable person would consider its answers facts or true.
We see this in Walters v. OpenAI, where a judge ruled for OpenAI after radio talk show host Mark Walters sued the tech giant for a ChatGPT output that falsely stated he was a defendant in an embezzlement lawsuit. The court held, among other actors, that Walters failed to produce sufficient evidence that OpenAI met the required level of fault under Georgia’s defamation laws. The court also noted that a reasonable reader wouldn’t necessarily believe ChatGPT’s response as factual information. ChatGPT and other AI companies warn their users via disclaimers that they may occasionally generate incorrect information.
For an AI hallucination to be defamatory, it must also be shared with a third party. Simply creating the incorrect output isn’t enough. However, disseminating the response doesn’t need to be widespread. Publication can occur even if the output is shared with just one other person, including the person who provided the prompt. In Walters v. OpenAI, the output was shared with at least one third party — the journalist who asked ChatGPT to summarize a court filing.
Can I Sue Over AI-Summarized Reviews of My Business?
On the other hand, if a GenAI tool produces a less-than-stellar output regarding your company, but scraped its response from actual reviews, your legal beef likely isn't with the AI company at all.
In this case, you’d have to take legal action against whoever wrote the review — not the review platform or the AI company. And you’d still have to prove the four elements of defamation, including that the review was factually inaccurate. A bad review isn’t illegal, no matter how much you want it to be.
But what if the output is sourced from fake reviews? AI-generated business summaries are only as accurate as the information they pull from. That is where the FTC’s Consumer Reviews and Testimonials Rule comes in. This October 2024 rule addresses the harm fraudulent reviews cause in several ways, including making it illegal to buy or sell fake reviews or testimonials, both positive and negative. It also applies to businesses that create misleading review websites that appear independent.
In this scenario, proving that an AI output based on a fake review is defamatory still requires showing the statement was false and harmed your business reputation. And the AI company most likely isn’t the liable party. As with a real review, you’d need to sue the person or entity that created or transacted the fake review, not the AI chatbot.
This area of law is evolving, and the legal questions surrounding AI-generated content are still developing. Getting legal advice from a defamation attorney is the best way to determine whether you have a viable defamation claim and avoid wasting your time (or money) on a fruitless lawsuit.
