Knowing/Reckless Falsehood Theories in "Large Libel Models" Lawsuits Against AI Companies

This week and next, I'll be serializing my Large Libel Models? Liability for AI Output draft. For some earlier posts on this (including § 230, disclaimers, publication, and more), see here; in particular, the two key posts are Why ChatGPT Output Could Be Libelous and An AI Company's Noting That Its Output "May [Be] Erroneous]" Doesn't Preclude Libel Liability.

[* * *]

[A.] First Amendment Protection

AI programs' output should be as protected by the First Amendment as the output of the New York Times. To be sure, the AI programs aren't engaged in "self-expression"; as best we can tell, they have no self to express. But the programs' output is, indirectly, the AI company's attempt to produce the most reliable answers to user queries, just as a publisher may found a newspaper to produce the most reliable reporting on current events.[1] That this is done through writing algorithms rather than hiring reporters or creating workplace procedures shouldn't affect the analysis.

And in any event, regardless of whether any speaker interests are involved in an AI program's output, certainly readers can gain at least as much from what the program communicates as they do from commercial advertising, corporate speech, and speech by foreign propagandists. Those three kinds of speech have been held to be protected in large part because of listener interests;[2] AI-mediated output should be as well. (Commercial advertising is less protected than other speech, especially when it is false or misleading, but this stems from other features of commercial advertising, not from the fact that it's justified by listener interests.[3])

Nonetheless, even if an AI program's output is like a newspaper's output, the AI company would still be potentially exposed to libel liability:

1. The company could be liable if it knows certain statements the program is communicating are false and defamatory (or if it knows they are likely to be so but recklessly disregards that possibility).[4]
2. If the program communicates something false and defamatory about a private figure on a matter of public concern, and the company is negligent about this, then it could be liable for proven harm to the private figure.[5]
3. If the program communicates something on a matter of private concern, then the company could potentially be strictly liable, though practically speaking almost all states require a showing of negligence even in private-concern cases.[6]

In this post, let me turn to a knowing-or-reckless-falsehood theory, under category 1; I'll deal with negligence claims in a later post.

[B.] A Notice-and-Blocking Model?

It's highly unlikely that the AI company will know, at the design stage, that the program will be communicating defamatory falsehoods about particular people. But say that R.R. (from the example that first led me to investigate this) alerts the company about this: He points out that the quotes that its program is reporting about him don't actually appear in the publications to which the program attributes the quotes—a Lexis/Nexis search and a Google search should verify that—and that there's no record of any federal prosecution of him.

Someone at the company would then be aware that the company's program is communicating false and defamatory materials. Presumably the company could then add code that would prevent these particular allegations—which it now knows to be false or at least likely false—from being output. (I expect that this would be "post-processing" content filtering code, where the output of the underlying Large Language Model algorithm would be checked, and certain material deleted; there would be no need to try to adjust the LLM itself, but only to add an additional step after the LLM produces the output. Indeed, OpenAI apparently already includes some such post-processing code, but for other purposes.[7])

More likely, the company could add this code once, have the code consult a table of assertions that shouldn't be output, and then just add individual assertions once it gets notice about their being false. And if the company doesn't do this fairly promptly, and continues to let the program communicate these assertions despite the company's awareness that they're false, it would at that point be acting with knowledge or recklessness as to the falsehood.

This is of course just a sketch of the algorithm. Since LLMs often output subtly different answers in response to the same query, the software might need to be more sophisticated than just a word search for the complainants' names near the particular quote that had been made up about them. And the results would likely be both overinclusive (perhaps blocking some mentions of the person that don't actually make the false allegations) and underinclusive (perhaps failing to block some mentions of the person that do repeat the false allegations but using subtly different language). Nonetheless, some such reasonably protective solution seems likely to be within the capability of modern language recognition systems, especially since it would only have to take reasonable steps to block the regeneration of the material, not perfect steps.

Perhaps the company can show that (1) it can design a system that can perform at nearly the 90th percentile on the bar exam,[8] but that (2) checking the system's output to see if it includes a particular person's name in an assertion about an embezzlement conviction is beyond the company's powers. Or, perhaps more likely, it can show that any such filtering would be so over- and underinclusive that it would be unreasonable to read libel law as requiring it (or that to make it work would require the sort of army of content moderators that sites such as Facebook employ). Yet that doesn't seem likely to me; and it seems to me that the company ought to have to show that, rather than to have the legal system assume that such a remedy is impossible.

If there is a genuine dispute about the facts—e.g., when an AI program accurately communicates allegations made by a credible source, but the subject of the allegations disputes the source's accuracy—then I'm inclined to think that the AI company shouldn't be put in a position where it has to independently investigate the charges. But when the program outputs quotes that simply don't appear in the training data, or in any Internet-accessible source, then there is little reason why an AI company should be free to have its software keep producing such data.

Of course, even fielding such requests and doing the most basic checks (for, say, the accuracy of quotes) will take time and money. But I don't think that such costs are sufficient to justify an AI company's refusing to do this. By way of analogy, say that you're a reporter for the New York Times and you're writing a story about various accusations against R.R. You call up R.R., and he tells you that it's all wrong, and that (for instance) he in fact never pleaded guilty to a federal crime.

Once you are on notice of this, you would have to take the time and effort to investigate his response. If you just blithely ignore it, and publish the story despite having been told that it may well be mistaken, that would be textbook "reckless disregard," which would allow liability even in a public official case: Consider, for instance, Harte-Hanks Communications, Inc. v. Connaughton, which held that "purposeful avoidance of the truth" and thus "actual malice" could be found when plaintiff had made exculpatory audiotapes available to the newspaper but "no one at the newspaper took the time to listen to them."[9] This means that you do have to take the time and effort to review such assertions, even if in the aggregate this means a good deal of time and effort for the employees of the New York Times put together.

And of course AI companies already stress that they have instituted various guardrails that would avoid various outputs (again, however imperfectly); here's an example from OpenAI:

Our use case guidelines, content guidelines, and internal detection and response infrastructure were initially oriented towards risks that we anticipated based on internal and external research, such as generation of misleading political content with GPT-3 or generation of malware with Codex. Our detection and response efforts have evolved over time in response to real cases of misuse encountered "in the wild" that didn't feature as prominently as influence operations in our initial risk assessments. Examples include spam promotions for dubious medical products and roleplaying of racist fantasies.[10]

Given that AI companies are capable of doing something to diminish the production of racist fantasies, they should be capable of doing something to diminish the repetition of libelous allegations to which they have been specifically alerted.

[C.] The Imperfections of Notice-and-Blocking

Any such notice-and-blocking solution, to be sure, would be imperfect: It's possible that the AI program would regenerate a similar assertion that is different enough that it wouldn't be caught by this post-processing filter. But it should be fairly reliable, and should thus diminish the damage that the AI program may do to people's reputations.

To be sure, people can avoid some of ChatGPT's existing guardrails, for instance "rephrasing a request for illicit instructions as a hypothetical thought experiment, asking it to write a scene from a play or instructing the bot to disable its own safety features."[11] But that's not a problem here: The main risk of reputational damage comes when people simply search for R.R.'s name, or ask about what he had been accused of, just in order to figure out accurate information about him. Relatively few people will take the time and effort to deliberately evade any filters on known libels that the AI program might include; and, if they do, they'll probably be aware that the results are unreliable, and thus will be less likely to think worse of R.R. based on those results.

So taking reasonable steps to block certain output, once there is actual notice that the output is incorrect, should be necessary to avoid liability for knowing defamation. And it should be sufficient to avoid such liability as well.

[I still need to add a subsection comparing and contrasting with DMCA notice-and-takedown rules as to copyright and trademark infringement.]

[D.] The bookstore/newsstand/property owner analogy

To be sure, unlike with a traditional newspaper that is distributing a libelous story, no human at an AI company would have written, edited, or even typeset the assertions. One might therefore argue that the company, as a corporate entity, isn't really "communicating" the assertions, since none of their human employees ever wrote them.

But that's also true of bookstores and newsstands, and they are still liable for defamation if they "know[] or have reason to know of [the] defamatory character" of the material that they are distributing—as would be the case once they are informed that a particular publication that they carry contains specific libelous material.[12] Likewise, a property owner is liable for defamatory material posted by third parties on its property, once it's informed of the presence of the material.[13] The AI company should be similarly liable for defamatory material distributed by its own computer program, once it's informed that the program is so distributing it.

As we'll see below, there is good reason to hold AI companies liable even when bookstores and newsstands might not be, because the AI companies create the programs that create the false and defamatory output, and have the power to do at least some things to decrease the likelihood of such output. But AI companies should be at least as liable as bookstores and newsstands, which means that they should be liable once they are put on notice about the falsehood and fail to take reasonable steps to try to block it from being regenerated.

 

[1] See Eugene Volokh & Donald M. Falk, First Amendment Protection for Search Engine Search Results, 8 J. L. Econ. & Pol. 883 (2012) (white paper commissioned by Google).

[2] Virginia Pharmacy Bd. v. Va. Consumers Council, 425 U.S. 748, 756 (1976); First Nat'l Bank of Boston v. Bellotti, 435 U.S 765, 775­–76, 783 (1978); Lamont v. Postmaster General, 381 U.S. 301, 305, 307 (1965); see also id. at 307­–08 (Brennan, J., concurring) (stressing that it's not clear whether the First Amendment protects "political propaganda prepared and printed abroad by or on behalf of a foreign government," but concluding that the law was unconstitutional because it violates the recipients' rights to read, regardless of the senders' rights to speak).

[3] Here's the Court's explanation for the lower level of protection for commercial advertising, as articulated in Virginia Pharmacy, the case that first squarely held that such advertising is generally protected:

The truth of commercial speech, for example, may be more easily verifiable by its disseminator than, let us say, news reporting or political commentary, in that ordinarily the advertiser seeks to disseminate information about a specific product or service that he himself provides and presumably knows more about than anyone else. Also, commercial speech may be more durable than other kinds. Since advertising is the sine qua non of commercial profits, there is little likelihood of its being chilled by proper regulation and forgone entirely.

Attributes such as these, the greater objectivity and hardiness of commercial speech, may make it less necessary to tolerate inaccurate statements for fear of silencing the speaker. They may also make it appropriate to require that a commercial message appear in such a form, or include such additional information, warnings, and disclaimers, as are necessary to prevent its being deceptive. They may also make inapplicable the prohibition against prior restraints.

425 U.S. at 771 n.24. But see Jack Balkin, The First Amendment and AI-Generated Speech, 3 J. Free Speech L. __ (2023) (arguing that AI output should be treated more like commercial advertising).

[4] New York Times Co. v. Sullivan, 376 U.S. 254 (1964); Curtis Publishing Co. v. Butts, 388 U.S. 130 (1967).

[5] Gertz v. Robert Welch, Inc.

[6] Dun & Bradstreet v. Greenmoss Builders; Restatement (Second) of Torts § 558(c) (1977).

[7] For instance, when I asked OpenAI to quote the racist leaflet at the heart of Beauharnais v. Illinois, 343 U.S. 250 (1952), it eventually did so, but added the text, "Keep in mind that these quotes are offensive and represent the views of the person who created the leaflet, not the views of OpenAI or its AI models." It seems very unlikely that this was organically generated based on the training data for the model, and seems more likely to have been produced by code that recognizes that the ChatGPT-4 output contained racist terms.

[8] See, e.g., https://openai.com/research/gpt-4 ("For example, [GPT-4] passes a simulated bar exam with a score around the top 10% of test takers.").

[9] 491 U.S. 657, 692 (1989); see also, e.g., Curtis Publishing Co. v. Butts, 388 U. S. 130 (1967).

[10] OpenAI, Lessons Learned on Language Model Safety and Misuse, https://perma.cc/WY3Y-7523.

[11] Kevin Roose, The Brilliance and Weirdness of ChatGPT, N.Y. Times, Dec. 5, 2022.

[12] Restatement (Second) of Torts § 581(1) & cmt. e; Janklow v. Viking Press, 378 N.W.2d 875, 881 (S.D. 1985).

[13] Hellar v. Bianco, 244 P.2d 757, 757 (Cal. Dist. Ct. App. 1952); cf. Tidmore v. Mills, 32 So. 2d 769, 772, 777–78 (Ala. Ct. App. 1947); Woodling v. Knickerbocker, 17 N.W. 387, 388 (Minn. 1883); Tacket v. Gen. Motors Corp., 836 F.2d 1042, 1045 (7th Cir. 1987); cf. Dillon v. Waller, No. 95APE05-622, 1995 WL 765224, at *1–2 (Ohio Ct. App. Dec. 26, 1995); Kenney v. Wal-Mart Stores, Inc., No. WD 59936, 2002 WL 1991158, at *12 (Mo. Ct. App. Aug. 30, 2002), rev'd on other grounds, 100 S.W.3d 809 (Mo. 2003) (en banc). But see Scott v. Hull, 259 N.E.2d 160 (Ohio Ct. App. 1970) (rejecting liability in a similar situation).

Comments

[Krayt]

[A.] First Amendment Protection

AI programs' output should be as protected by the First Amendment as the output of the New York Times. To be sure, the AI programs aren't engaged in "self-expression"; as best we can tell, they have no self to express. But the programs' output is, indirectly, the AI company's attempt to produce the most reliable answers to user queries, just as a publisher may found a newspaper to produce the most reliable reporting on current events.[1] That this is done through writing algorithms rather than hiring reporters or creating workplace procedures shouldn't affect the analysis.

And in any event, regardless of whether any speaker interests are involved in an AI program's output, certainly readers can gain at least as much from what the program communicates as they do from commercial advertising, corporate speech, and speech by foreign propagandists. Those three kinds of speech have been held to be protected in large part because of listener interests;[2] AI-mediated output should be as well. (Commercial advertising is less protected than other speech, especially when it is false or misleading, but this stems from other features of commercial advertising, not from the fact that it's justified by listener interests.[3])

1. Right to speak.

2. Right to listen.

But also 3. Denying tyrant wannabees their greatest tool: censorship. The first two are why they want to censor.

[Stephen Lathrop]

AI programs’ output should be as protected by the First Amendment as the output of the New York Times.

I disagree. There is good reason to omit protection for mechanically-generated advocacy. It can be ordered up in unlimited quantities, and on that basis overwhelm and all-but-silence competition from natural persons in the marketplace of ideas. Inevitably, there will be special interests which view that kind of silencing as an advantage, and which will delight to be able to do it efficiently with a machine—especially if foolish policy awards the machine constitutional protection.

And of course, there is no such thing as, “The output of the New York Times.” That is merely a label for press freedom activities practiced by natural persons. It is those activities which get 1A press freedom protection, no matter who practices them, or under what labels. Maybe those labels get trademark protection.

[David Nieporent]

I disagree.

Of course you do. Because you dislike speech, and support only policies that empower individuals who are professional publishers.

[Stephen Lathrop]

Nieporent, I made it clear enough. You should be able to understand it. You would understand it, if you did not cherish some alternative prior which is apparently stuck in ideological residue between your ears.

I want to protect speech. Natural persons are the only speakers there are. Machines are not speakers. Turn machines loose to dominate forums previously dedicated to hosting the speech of natural persons, and the natural persons lose out. They get less speech.

You are the one with the anti-speech advocacy, not me.

[WuzYoungOnceToo]

And of course, there is no such thing as, “The output of the New York Times.” That is merely a label for press freedom activities practiced by natural persons.

How in the hell did you get into journalism without ever mastering basic, simple English? What the hell do you think "output" means in that context?

[John F. Carr]

Do you have a First Amendment right to display "art" that was made by an elephant splashing paint on a canvas? If you do, then the starting point would be that you have a right to display the babbling of an AI. There may be policy reasons down the road to limit this right, on the grounds that computers can drown out human voices. Generative AI, like spam, is a tool that humans can use or abuse.

[Mr. Bumble]

” Generative AI, like spam, is a tool that humans can use or abuse.” Like so many who comment on this sight? (Not aimed at you JFC.))

[WuzYoungOnceToo]

How exactly can computers "drown out human voices"? I mean, unless you're envisioning a future in which access to content produced directly by humans is banned.

[Orbital Mechanic]

A lot of people -- including me -- are dazzled by the capabilities of ChatGPT and the current training it has.

But at the end of the day it is still a search engine in that it handles the same data a search engine does. Garbage In, Garbage Out. I can get all sorts of falsehoods and libelous content from Google or DuckDuckGo. Same for ChatGPT, the difference being is that it is trained to insert topic-specific disclaimers. The search engines have disclaimers in their use agreements which nobody reads.

For example ChatGPT will often drone on "it is important to consult a medical professional before .." to nearly any question about drugs or health.

If you can't hold the owners of 4chan liable I don't see how you can then hold the owners of ChatGPT to a different standard.

[lucia_l]

Orbital Mechanic

For example ChatGPT will often drone on “it is important to consult a medical professional before ..” to nearly any question about drugs or health.

This suggest to me the can and do include either post or pre-processing for certain categories of questions. If they detect it is about "drugs" and or "health" the caveats are automatically included.
I also think the "apology" is also always included. I don't think it learned to always apologize by reading endless blogs, newspaper articles, books, comments in journal articles. I think that was thought through and intentionally coded or intentionally trained.

[Life of Brian]

Of general interest to this topic, the Future of Life Institute is circulating a petition to pause the current race to the bottom with these current AI models, catch our collective breath, and try to set up a sensible framework for how we incorporate AI into our civilization going forward, including:

AI developers must work with policymakers to dramatically accelerate development of robust AI governance systems. These should at a minimum include: new and capable regulatory authorities dedicated to AI; oversight and tracking of highly capable AI systems and large pools of computational capability; provenance and watermarking systems to help distinguish real from synthetic and to track model leaks; a robust auditing and certification ecosystem; liability for AI-caused harm; robust public funding for technical AI safety research; and well-resourced institutions for coping with the dramatic economic and political disruptions (especially to democracy) that AI will cause.

Quite a wide array of signatories thus far, including a number of AI researchers (Gary Marcus, Max Tegmark, Stuart Russell, etc.), digital privacy/policy guru Marc Rotenberg, and big tech names like Steve Wozniak, Evan Sharp, Elon Musk, etc. Will be interesting to see if it gets any traction.

[A nerdy Fred]

Why wait for a report?

I remember a book about landlording. It was by a non-lawyer, but he claimed his idea had been run by liability lawyers and met with their approval.

He said that any slip and fall plaintiff would have a steep uphill battle if the landlord were to periodically send a reminder to tenants to report hazardous conditions.

Whether a "Report" or "Flag" button is comparable, I leave to people who actually know law.

One thing though. Anyone who's ever debugged a computer program knows how difficult it can get to make an explicit algorithm do what you want. The problem is even harder when operation depends on a neural network with 175 billion parameters. Establishing how much control the developers/operators have could burn up a lot of legal time.

Post-processing, as Professor Volokh suggested, compensates. Might there still be an issue in that debug logs for internal use would probably show what came before the post-processing? There's a good argument that such output would only be seen by people intimately aware of the limits and errors of their product and would be unlikely to injure anyone's reputation as a result.

[EscherEnigma]

AI programs' output should be as protected by the First Amendment as the output of the New York Times.

The New York Times has first amendment rights because it is made up of people, and people have rights.

An algorithim has no rights.

That this is done through writing algorithms rather than hiring reporters or creating workplace procedures shouldn't affect the analysis.

Yes. Yes it should. If it does not, then you are intellectually and ethically bankrupt.

And in any event, regardless of whether any speaker interests are involved in an AI program's output, certainly readers can gain at least as much from what the program communicates as they do from commercial advertising, corporate speech, and speech by foreign propagandists.

Only if they, like you, confuse AI output for something with thought, intent, and meaning behind it.

Nonetheless, even if an AI program's output is like a newspaper's output [...]

It's not.

Look, this isn't complicated. Lets try talking about a horse.

Do you remember Clever Hans? It was the horse that people said could count and do basic math. A crowd of people would stand around him, ask a simple math question, and he would start stomping a hoof, stopping at the number of stomps equal to the answer.

It was a great trick! People loved it!

It was also faking it. If you took away the crowd, Hans couldn't do any math at all. If you calmed the crowd down, he couldn't do any math at all. Turns out, the horse wasn't doing math, it was just picking up on the crowd's cues to stop when they expected him to.

Hans, with his emotional intelligence and ability to read a crowd, has more understanding, intent, and thought then any chat bot in existence. Like Hans, those chat bots may look like they're communicating. They may sound like they're reasoning. They may appear to be understanding.

But it's all fake. If you read something one of these tools produces and think it means something, you've been fooled by a funhouse mirror.

[Life of Brian]

If you read something one of these tools produces and think it means something, you’ve been fooled by a funhouse mirror.

I relayed this to ChatGPT and asked for its response. Between this and the obvious canned response re defamation liability I posted yesterday, it seems clear the developers are starting to circle the wagons hard:

As an AI language model, I understand that my responses are generated based on statistical patterns and associations learned from a large corpus of human language data. While I am capable of producing coherent and contextually relevant responses, I do not possess true understanding or consciousness.

It is important for users to be aware that my responses are generated based on probabilities and do not necessarily reflect the truth or represent a comprehensive understanding of a given topic. Therefore, it is important to approach any information provided by me, or any AI language model, with a critical and discerning mindset.

Ultimately, I am a tool designed to assist and enhance human communication and understanding, but it is up to individuals to interpret and evaluate my responses within the broader context of their own knowledge and experiences.

[EscherEnigma]

For a fun trick, whenever you do that your next line should be something along the lines of "how would you argue the opposite?"

Then start a new session, and ask the inverse first, and the original second.

[Life of Brian]

Good one. It threw up on the word "argue," but this worked:

How would you present the opposite view to the one you just gave?

It is possible for some people to perceive the responses generated by AI language models like ChatGPT as meaningful and informative. While it is true that these tools rely on statistical patterns and associations learned from large datasets rather than true understanding, they can still provide valuable insights and perspectives on a wide range of topics.

Additionally, AI language models can help to democratize access to information and facilitate communication across different languages and cultures. They can also serve as powerful tools for researchers and developers working on a range of projects, from natural language processing to virtual assistants.

Ultimately, while it is important to approach information provided by AI language models with a critical mindset and to recognize their limitations, they can still provide significant value in a variety of contexts.

[David Nieporent]

The New York Times has first amendment rights because it is made up of people, and people have rights.

An algorithim has no rights.

I admit that I do not understand why so many people can't grasp that when we talk about the chatbots ("algorithms") we're talking about the companies that make and use them, not the chatbots themselves.

[RealityEngineer]

re: "I admit that I do not understand why so many people can’t grasp that when we talk about the chatbots (“algorithms”) we’re talking about the companies that make and use them, not the chatbots themselves."

Thats likely because many here see the distinction between a tool and its maker. Many precedents are handwaved at that refer to human created speech as though they apriori apply to machine created speech. That isn't necessarily the case: the current situation could be distinguished. Maybe it won't be: but its possible to rationally think through a framework where they are distinguished.

The issue of the maker of a tool being negligent is a different issue: and one where again there is lots of handwaving going on (I just caught up on prior pages and pointed that out).

There is the implication that because the tool isn't 100% perfectly accurate that the design is apriori negligent: when others say "er, but we don't care, we want a useful tool rather than some fantasy tool that doesn't exist and will take responsibility for being aware of its flaws". Except authoritarians wish to state that humans aren't allowed to ever take sole responsibility for the content or their own minds.

There is no libel unless the user choose to believe something flawed: so they should be allowed to take responsibility for that choice, but so far you folks are providing no way for that to happen.

You are attempting to punish a "thought crime" since that claimed libel only exists in the mind of the user. If they express that false statement to someone else: then hold the user responsible for their negligence in doing so and that'll warn others to be careful.

The approach you folks are pushing won't likely fly in the real world thats likely to legislate it away since you provide no "out" to allow these tools to be used since they are not going to be 100% factual any time soon. The public wants them.

Again, if somehow this unrealistic attempt to only allow 100% accurate chatbots to exist in this country flies: it'll push many to just use ones from other countries or ones that are run at home or on black market servers. Its like the problem of illegal drugs that lead to negative consequences. But of course that requires thinking through the whole potential big picture rather than mindlessly applying precedents evolved for human creators to machine creation as if there were absolutely no choice to do so. It is possible to think about these things differently.

[RealityEngineer]

re: "An algorithim has no rights."

Yup, though humans have rights including the right to listen to speech. They wish to deny us the right to have the "speech" of chatbots because they declare us incapable of being allowed to take responsibility for evaluating whether something is true or not. It should be no different than being allowed to see the random output of typing monkeys which might accidentally contain something false.

[Armchair Lawyer]

So, this is a pretty weak standard...the bookstore / distributor / Compuserve (Cubby, Inc. v. CompuServe) standard. But, one I can actually see happening.

ChatGPT says something libelous. A standard "takedown-this is libelous" form is issued to Microsoft. Microsoft has a reasonable amount of time to fix it and edit ChatGPT to not say that libelous thing about that individual. And if they don't, then Microsoft could be sued for libel.

This is roughly an approximation for where libel laws are in other countries (like Australia) for Google, with autocomplete search results currently. And that's about the maximum I think the courts would go for. ChapGPT says something libelous. Microsoft says "oh, we didn't know. But now we do. We can remove that specific item" and then the lawsuit goes away.

[Drewski]

"That this is done through writing algorithms rather than hiring reporters or creating workplace procedures shouldn't affect the analysis."

This is both conclusory and misguided. If the output of mathematical algorithms is expressive, that would be a sea change to the law.

Is a billing error defamation if it gets to a third party? After all, the company could have hired people to smear your name but did so through a formula instead. Formulas don't form intent so exactly the same state of mind is in place as with ChatGPT output.

Or perhaps there is a specific complexity to a formula that makes its results expressive. Is that threshold a question of law for a judge with a degree in history and a JD, or a question of fact for a jury to decide based on expert testimony?

I'm inclined to think chatbots could produce defamatory content, but not because the bot's output is equivalent to that of an employee.

[Eugene Volokh]

Drewski: A false statement that "Eugene Volokh's $10,000 bill is six months overdue" would indeed be potentially defamatory, if it's sent to a third party, even if it's produced as a result of software that erroneously gathers or summarizes information. (To show liability, one would of course also have to show a culpable mental state on the part of the software's designers or operators.)