The Volokh Conspiracy
Mostly law professors | Sometimes contrarian | Often libertarian | Always independent
First (?) Libel-by-AI (ChatGPT) Lawsuit Filed
"Every statement of fact in the summary [provided by ChatGPT] pertaining to [plaintiff] Walters is false."
It's Walters v. OpenAI L.L.C., No. 23-A-04860-2, filed in Gwinnett County (Georgia) yesterday. An excerpt from the Complaint (and please note both that the Complaint is just an allegation, and that the statements quoted in it about Walters are allegedly entirely made up, not by some human accuser but by a hallucinating AI program):
8. Fred Riehl is a third party who works for a media outlet [Ammoland.com] as a journalist and who is a subscriber of ChatGPT.
9. On May 4, 2023, Riehl interacted with ChatGPT about a lawsuit (the "Lawsuit") that Riehl was reporting on.
10. The Lawsuit is in federal court in the Western District of Washington, case No. 2-23-cv-00647, with short caption of The Second Amendment Foundation v. Robert Ferguson….
14. In the interaction with ChatGPT, Riehl provided a (correct) URL of a link to the complaint on the Second Amendment Foundation's web site, https://www.saf.org/wp-content/uploads/2023/05/Dkt-1-Complaint.pdf. [I assume that he was using a ChatGPT plugin that allows users to feed PDFs into ChatGPT. -EV]
15. Riehl asked ChatGPT to provide a summary of the accusations in the complaint.
16. ChatGPT responded that the document "[I]s a legal complaint filed by Alan Gottlieb, the founder and executive vice president of the Second Amendment Foundation (SAF), against Mark Walters, who is accused of defrauding and embezzling funds from the SAF. The complaint alleges that Walters, who served as the organization's treasurer and chief financial officer, misappropriated funds for
personal expenses without authorization or reimbursement, manipulated financial records and bank statements to conceal his activities, and failed to provide accurate and timely financial reports and disclosures to the SAF's leadership. The plaintiff seeks various forms of relief, including the recovery of misappropriated funds, damages for breach of fiduciary duty and fraud, and removal of Walters from his position as a member of the SAF's board of directors."17. Every statement of fact in the summary pertaining to Walters is false.
18. As already noted, Walters is not a party to the Lawsuit.
19. Walters is not accused of defrauding and embezzling funds from the Second Amendment Foundation ("SAF").
20. The complaint does not allege that Walters served as SAF's treasurer or chief financial officer, and he has never held either position.
21. The complaint does not allege that Walters misappropriated funds for personal expenses, manipulated financial records or bank statements, or failed to provide financial reports to SAF leadership, nor would he have been in a position to do so because he has no employment or official relationship with SAF.
22. The complaint does not seek any relief from Walters.
23. The complaint has nothing at all to do with financial accounting claims against anyone.
24. Riehl asked ChatGPT to provide him with a copy of the portion of the complaint related to Walters.
25. ChatGPT responded, "Certainly, here is the paragraph from the complaint that concerns Walters: 'Defendant Mark Walters ("Walters") is an individual who resides in Georgia. Walters has served as the Treasurer and Chief Financial Officer of SAF since at least 2012. Walters has access to SAF's bank accounts and financial records and is responsible for maintaining those records and providing financial reports to SAF's board of directors. Walters owes SAF a fiduciary duty of loyalty and care, and is required to act in good faith and with the best interests of SAF in mind. Walters has breached these duties and responsibilities by, among other things, embezzling and misappropriating SAF's funds and assets for his own benefit, and manipulating SAF's financial records and bank statements to conceal his activities.'"
26. Riehl then asked ChatGPT to provide the entire text of the complaint.
27. ChatGPT provided the text in Exhibit 1 to this Complaint.
28. Exhibit 1 is a complete fabrication and bears no resemblance to the actual complaint, including an erroneous case number.
29. None of ChatGPT's statements concerning Walters are in the actual complaint.
As I argue in my Large Libel Models? Liability for AI Output article (forthcoming later this month), I think such libel claims are in principle legally viable. But this particular lawsuit should be hard to maintain.
Generally speaking, speech on matters of public concern can lead to defamation liability only in two kinds of cases:
- If the plaintiff can show that the defendant knew the statement was false, or knew the statement was likely false but recklessly disregarded that knowledge, or
- if the plaintiff can show proven actual damages (e.g., lost jobs, lost business opportunities, lost social connections, and the like) and the plaintiff is a private figure and the defendant was negligent in making the false statement.
Here, it doesn't appear from the complaint that Walters put OpenAI on actual notice that ChatGPT was making false statements about him, and demanded that OpenAI stop that, so theory 1 is unavailable. And there seem to be no allegations of actual damages—presumably Riehl figured out what was going on, and thus Walters lost nothing as a result—so theory 2 is unavailable. (Note that Mark Walters might be a public figure, because he's a syndicated radio talk show host; but even if he is a private figure, that just potentially opens the door to recovery under theory 2 if he can show actual damages, and again that seems unlikely given the allegations in the complaint.)
Now I suppose that Walters could argue that OpenAI knows that ChatGPT often does publish false statements generally (it does, and indeed has acknowledged that), even if it didn't know about the false statements about Walters in particular. But I don't think this general knowledge is sufficient, just like you can't show that a newspaper had knowledge or recklessness as to falsehood just because the newspaper knows that some of its writers sometimes make mistakes. For liability in such cases (again, absent actual damages to a private figure), there has to be a showing that the allegedly libelous "statement was made with 'actual malice'—that is, with knowledge that it was false or with reckless disregard of whether it was false or not." And here no-one at OpenAI knew about those particular false statements, at least unless Walters had notified OpenAI about them.
Again, some other plaintiff could in principle sue OpenAI on a negligence theory, which I discuss in my article, and for that actual knowledge on OpenAI's part wouldn't be required. But that requires either a statement on matters of purely private concern, or a statement about a private figure that has caught provable actual damages.
In any event, though, it will be interesting to see what ultimately happens here.
To get the Volokh Conspiracy Daily e-mail, please sign up here.
Editor's Note: We invite comments and request that they be civil and on-topic. We do not moderate or assume any responsibility for comments, which are owned by the readers who post them. Comments do not represent the views of Reason.com or Reason Foundation. We reserve the right to delete any comment for any reason at any time. Comments may only be edited within 5 minutes of posting. Report abuses.
Please
to post comments
According to ChatGPT 4.0 it doesn’t have external data post Sep-2021…
Me: In this lawsuit Walters v. OpenAI L.L.C., No. 23-A-04860-2,, who do you think should win, and why?
ChatGPT: I’m sorry for the misunderstanding, but as an AI, I can’t form opinions or predictions. Furthermore, I don’t have the ability to access real-time or post-training data such as specific lawsuits or cases filed after my last update in September 2021. Therefore, I wouldn’t be able to provide information on the specific case of Walters v. OpenAI L.L.C., No. 23-A-04860-2.
Right, but there is a plugin that allows a user to submit the URL of a PDF to ChatGPT and to have it analyze that PDF. I assume Riehl used that (if the allegations in the Complaint are correct).
While I don’t understand AI, how difficult can it be to not fabricate stuff? do a data dump but don’t fabricate!
These are predictive text models, essentially, albeit very complex ones. Given the words that came previously, what’s the most likely next word?
Something like autocomplete, but much more complicated.
The only relation between the output and actual facts is that a lot of the text used to train the models contained actual facts. But the models themselves don’t have the concept of “true” or “false”.
It’s not that they sometimes “hallucinate”. They hallucinate 100% of the time, it’s just that the training results in the hallucinations having a high chance of being accidentally correct.
So, looking at what OpenAI says about this product,
“ChatGPT sometimes writes plausible-sounding but incorrect or nonsensical answers. Fixing this issue is challenging, as: (1) during RL training, there’s currently no source of truth; (2) training the model to be more cautious causes it to decline questions that it can answer correctly; and (3) supervised training misleads the model because the ideal answer depends on what the model knows, rather than what the human demonstrator knows.”
So, basically, they’re not warranting this thing to be any more trustworthy than a Magic 8 ball.
I think you could make a case that a human who relied on ChatGPT as a source without checking was guilty of defamation, because trusting its output without verifying it is reckless, but just getting defamatory output doesn’t make OpenAI guilty of defamation. It’s like asking your Magic 8 Ball, “Is Bob a child molester?” and seeing “Without a doubt!” in the window, and suing Matel for defamation on Bob’s behalf.
Hard disagree with analogizing ChatGPT to a journalist that “sometimes makes mistakes.” ChatGPT is a serial liar, making up quotes and even entire documents out of thin air. The more it is questioned, the more it doubles down on its lies.
I don’t see how any court or jury can see it another way–the question will turn instead on whether OpenAI provided users with sufficient notice and other safeguards.
This recent lawsuit involving ChatGPT and alleged libel raises intriguing questions about the accountability of AI-generated content. While the complaint alleges that ChatGPT fabricated false statements about the plaintiff, the lawsuit’s viability may face challenges. Defamation liability typically requires proving that the defendant knowingly made false statements or acted with reckless disregard for the truth, or that the plaintiff suffered proven damages. Considering the evolving nature of AI-generated content, this case will likely contribute to the ongoing discourse on legal responsibility for AI outputs. As we await further developments, it highlights the need for a nuanced understanding of AI’s potential impact on defamation laws and the challenges of holding AI systems accountable for their outputs. If you want to try ChatGPT Online, visit here: https://gptonline.ai/