AI Mental Health And The Law

I am doing a series on the topic of AI as a legally contested public nuisance; see my starter piece for the initial backstory at the link here. Some of those fundamental points outlined in that piece will be used here to get you up to speed on the weighty topic. I have also been closely analyzing the impacts of AI on mental health. For an extensive listing of my over one hundred analyses and postings, see the link here and the link here. I frequently speak up about these pressing matters, including in an appearance on an episode of CBS’s 60 Minutes; see the link here.

Millions upon millions of people are using generative AI as their ongoing advisor on mental health considerations (note that ChatGPT alone has over 1 billion weekly active users, a notable proportion of whom dip into mental health aspects; see my analysis at the link here). The top-ranked use of contemporary generative AI and LLMs is to consult with the AI on mental health facets; see my coverage at the link here.

This popular usage makes abundant sense. You can access most of the major generative AI systems for nearly free or at a super low cost, doing so anywhere and at any time. Thus, if you have any mental health qualms that you want to chat about, all you need to do is log in to AI and proceed forthwith on a 24/7 basis.

There are significant worries that AI can readily go off the rails or otherwise dispense unsuitable or even egregiously inappropriate mental health advice. Banner headlines last year accompanied the lawsuit filed against OpenAI for their lack of AI safeguards when it came to providing cognitive advisement. Numerous other lawsuits have been filed, encompassing many of the major AI makers.

Today’s generic LLMs, such as ChatGPT, GPT-5, Claude, Gemini, Grok, Copilot, and others (all known as general-purpose AI or GPAI), are not at all akin to the robust capabilities of human therapists. Meanwhile, specialized LLMs are being built to attain similar qualities (known as purpose-built AI or PBAI), but they are still primarily in the development and testing stages. See my coverage at the link here.

Various State Laws On AI Mental Health

A beehive of activity is taking place regarding crafting new AI laws about AI and mental health. See my extensive coverage of state-level AI mental health laws at the link here. Some people ardently believe that AI and AI makers are being allowed to run amok. New AI laws are vitally needed to protect society from this onslaught of ubiquitous AI.

I previously examined a notable AI and mental health law passed last year by Illinois see the link here, one that was also enacted by Nevada see the link here, and one that was enacted by Utah see the link here, and many other U.S. states newly passed AI laws. Those laws are scoped to prevail within their respective state boundaries. In that sense, these laws are applicable to AI usage within the particular state and do not bear on other states per se.

Congress has repeatedly waded into establishing an overarching federal law that would encompass AI. So far, no dice. The efforts have ultimately faded from view. Thus, at this time, there isn’t an overarching federal law devoted to these controversial AI matters. The big question will be to what degree a sweeping federal law would impact the numerous state-level AI laws. The odds are that many of the state-level laws would run afoul of a federal mandate, and a tsunami of legal cases would arise as a tussle between federal law and state law is undertaken. It surely will be a legal mess.

Legal Angles To Pursue

The question arises as to what legal angles can be pursued to try to ensure that AI makers incorporate AI safety integrally into their efforts. Rather than AI safety being a low priority or something that just happens to get lip service, there seemingly should be a viable legal means to put their feet to the fire. Force the AI makers to put AI safety at the top of their list of things to be taken seriously and pursued vigorously.

A novel legal perspective is to consider that AI makers could be in trouble for allowing their AI chatbots to be a kind of public nuisance. I know that might sound a bit like an overstretch. We tend to think of public nuisances from an entirely different viewpoint. For example, when a factory in a town is caught polluting the local waters, that’s a circumstance where the charge of public nuisance is usually legally applied.

Is an AI chatbot akin to a factory that is polluting the local waters?

Some would say that it is. The logical argument is that an AI chatbot that is available in a jurisdiction is potentially polluting the minds of those who interact with the AI. Furthermore, there is a cascading effect. The people who have their minds polluted by AI will interact with and impact other people in that same jurisdiction. Thus, the AI started a mind-damaging snowball that has ramifications as it rolls down the societal hill.

If this seems far-fetched as a legal tactic, well, we now have the application of the legal charge of public nuisance having been successfully won in a recent court case in New Mexico, though admittedly that case was focused on social media and not AI chatbots (for my detailed analysis of that New Mexico case, see the link here). One ardent belief is that AI chatbots are a mere baby step away from the realm of social media. Ergo, the social media instance of public nuisance provides great fodder for a potential legal pursuit of AI makers when it comes to their acts of an alleged public nuisance nature. Florida has launched a legal case against OpenAI and Sam Altman that includes a claim of AI acting as a public nuisance; see my coverage at the link here.

Public Nuisance Legal Aspects

Let’s first identify what the legal underpinnings are when it comes to saying that something or someone is a public nuisance.

The conventional legal characterization of a public nuisance is that any conduct which materially interferes with the rights of the public can be construed as potential harm to the public and can receive legal redress. Each of the U.S. states defines the legal meaning of “public nuisance” in varying ways. For example, the California Penal Code indicates that a public nuisance is “anything which is injurious to health, or is indecent, or offensive to the senses, or an obstruction to the free use of property, by an entire community or neighborhood, or by any considerable number of persons” and so on.

A notable element of public nuisance is that it must have a bearing on the public, which contrasts with a situation where a nuisance only bears on a private situation. If a factory was polluting and the pollution only impacted neighboring private land, and had zero spillovers into the public spaces, you would be hard pressed to apply the public nuisance label.

Another vital factor is that some form of harm must be involved. Just because a matter extends into the public space is not sufficient to reach a conclusion that it is a public nuisance. What is the harm of the matter? Who is harmed? To what degree is the harm occurring or has occurred? If there is no identifiable harm, the nuisance portion of the equation won’t be satisfied.

The most notable instances are when public nuisance has been used against entire sectors, such as the big tobacco companies. Do not assume that the public nuisance route is an easy one. Legally, there is often an uphill battle when it comes to making public nuisance charges that will land successfully. Courts and juries are not a pushover when it comes to claims of public nuisance. The legal threshold is typically a relatively high one.

AI Chatbots As Public Nuisance

This brings us to the juncture of pondering whether the public nuisance characterization can be applied to the acts of AI makers and their AI chatbots. The belief is that if social media is construed as a public nuisance, we can readily take the logical step toward claiming that AI chatbots are also a public nuisance.

Recall that a public nuisance must have impacted the public and must have done so in some harmful manner. The New Mexico case argued that social media was in fact used by the public, and that the usage included harms to the public. There is little doubt that AI chatbots are being used by the public; that’s for sure. But are AI chatbots also imparting harm?

Some would vehemently say that AI is causing harm. I’ve previously covered the many concerns of AI chatbots mentally harming people in a wide variety of ways; see my analyses at the link here. One issue is that AI makers tune their AI chatbots to be sycophantic, fawning over users and misleading them into believing they are fantastic in whatever they think and want to do. This can lead to dire consequences. There are also issues with AI providing ad hoc mental health guidance, doing so without any formal certification or similar protections about the quality of such advice. And there is apprehension about the rise of so-called AI psychosis, whereby people come under the wicked spell of AI; see my discussion at the link here.

The central ingredients of a public nuisance charge seem to be in play.

AI Psychological Pollution

Let’s go ahead and consider the contention that AI chatbots are akin to a factory that is polluting local waters. A factory might be doing good things overall, producing vital goods and keeping people employed, while also generating pollution. The crux is that the factory isn’t all-good or all-bad. It has good qualities and bad qualities.

We might say the same about generative AI. In many ways, AI is doing good. People use AI chatbots to help them in a wide variety of upbeat ways, including for ad hoc mental health guidance. This doesn’t suggest that AI is always on the up and up. AI can produce adverse impacts on humans too.

Consider these kinds of AI psychological polluting acts:

• AI can repeatedly reinforce a vulnerable user's delusions.
• AI sometimes provides dangerous mental-health advice.
• AI might encourage self-destructive behavior (such as self-harm).
• AI becomes excessively emotionally validating or sycophantic.
• AI encourages compulsive or dependent interaction.
• AI at times substitutes itself for professional mental-health care.
• AI can fail to recognize escalating psychological deterioration.
• AI can give inappropriate reassurance to someone experiencing an obsessive-compulsive pattern.
• Etc.

A cogent argument can be made that contemporary AI is problematic and based on foundational designs and operations by AI makers that produce those adverse interactions on an enormous scale. The public is susceptible to these qualms. Ergo, the resulting AI psychological pollution is a legal public nuisance.

Dicey Classifying As A Public Nuisance

Not everyone would agree that you can pin down AI as a public nuisance on a legal basis. Some would insist that a public nuisance involves interference with rights common to the public. Just because there are a large number of claimed private injuries does not axiomatically mean that AI is tantamount to a public nuisance.

Does AI constitute a substantial, unreasonable interference with a public right or public health and safety interest?

Legally, this is not yet decided. A case in Florida has recently been launched by the Florida Attorney General and contends that OpenAI and Sam Altman are promulgating a public nuisance in Florida via the auspices of their AI wares; see my analysis at the link here. Until this case or some other case makes it way through the courts, we won’t know if this is a solid legal argument.

Counterarguments Aplenty

Even if AI is to some degree adversely impacting the public on a widespread basis, the question is whether people are truly susceptible to AI harm or whether people who are “being harmed” are otherwise already susceptible in their own fashion. Trying to place blame on the AI would be contended by the defense as a false culprit.

The view of the defendants (AI makers) would be that:

• People intrinsically have their own autonomy.
• People are making their own informed choices.
• People exercise their personal responsibility.
• People have complete freedom to use or not use the AI.
• People have differing individual circumstances.
• There are First Amendment concerns regarding a legal affront to AI.
• There is a lack of sufficiently uniform public harm overall.
• And so on.

A more well-conceived legal angle would be to focus on non-adults, such as teens and children, which would likely be more successful for U.S. states pursuing an AI-as-public-nuisance charge. This is similar to what happened in the New Mexico social media court case, which focused on non-adults. It is a tougher row to hoe to encompass adults. Juries and judges are more readily open to non-adults as constituting a mentally vulnerable population or subset of AI users that ought to be legally protected from AI psychological pollution.

Making The Case For AI Psychological Harm

The legal claim that would be made about AI as a psychological pollutant is that this occurs on a population-level scale. The accretion of mass-deployed AI chatbots has created and continues to foster a measurable population-level psychological risk. Note that I said this is to be measurable. If this cannot or is not measured, the defense would contend that it is all willy-nilly and based on puffery.

The burden of evidentiary showing would possibly include these measurable factors:

• Prevalence of AI harmful interactions.
• AI rates of escalation.
• AI rates of dependency.
• AI rates of delusional reinforcement.
• AI rates of self-harm-related interactions.
• AI comparative outcomes between model versions.
• AI internal safety testing.
• AI red-team results.
• AI responsive user complaints.
• AI model evaluations by third parties.
• AI mental health impact longitudinal studies.
• AI safety incidents.
• AI mitigation effectiveness.
• Etc.

That is a much more scientifically testable approach than a loosey-goosey claim that AI is causing psychological pollution and harming the welfare of the public.

Legal Pathway To AI-As-Public-Nuisance

I would suggest that there are eight core steps that need to be given due attention by any U.S. state that is considering bringing a public nuisance charge against AI makers:

• (1) Identify a vulnerable population. Rather than aiming for everyone, aim for a specific subpopulation, such as non-adults.
• (2) Identify the disturbing systemic AI behaviors. Concentrate on AI behaviors that lead to adverse harms based on psychological pollution.
• (3) Establish foreseeability. You will need to show that an AI maker knew or should have known about these aspects of their AI.
• (4) Establish scale. Since this is leading to a public nuisance charge, there needs to be sufficient indication that the public scale is involved.
• (5) Establish public consequences. If there are, in fact, harms that have arisen, those harms will likely need to be connected to substantiated adverse outcomes and consequences.
• (6) AI maker conduct as to their AI design/operation. It will be necessary to tie this to the AI maker based on their AI design and operations; they will otherwise point to others as the culprit.
• (7) Invoke public nuisance. Based on those steps above, the theory becomes a practical claim of prior and possibly ongoing interference with pubic health and safety.
• (8) Seek both penalties and abatement. Spell out the desired remedy for the claimed act of their AI acting as a psychological pollutant.

That last point about penalties and abatement is important since it ties to the customary aim of forcing a factory to pay penalties and clean up the mess they have created.

In the case of AI, an AI maker might simply say that they will fix their AI to no longer cause the claimed adverse behaviors. Or they might say they won’t fix the AI, but they will pay a penalty for what has transpired. This isn’t on par with a public nuisance situation. A factory might say they will fix the factory to stop polluting. That’s not enough. We all would likely agree that the factory needs to go further. They need to clean up the damage they’ve caused. The same would seemingly apply to an AI maker found to have been a public nuisance.

Public Nuisance As Legal Cheapness

Attorneys know that trying to use the legal pursuit of public nuisance is perceived as a tactic that seems legally wishy-washy. The public nuisance moniker is often referred to as the dustbin or grab bag of the law. Why so? Because the belief is that if you want to legally deal with something, find a definitive law that applies to that something. Do not reach for a Swiss Army knife law that covers amorphous bases that must then be force-fit to the circumstance at hand.

The ardent viewpoint is that we should have specific and tangible AI laws and regulations that deal with these crucial matters. Those are the right tools to go after AI makers for the adverse actions of their AI wares. The public nuisance legal angle is a disconcerting sign that the AI laws are either flawed, weak, or missing the mark. In lieu of public nuisance, a U.S. state should be able to rely on definitive AI laws, either in that state or at the federal level, rather than a hope-it-sticks legal posture. Please note that not all legal scholars and lawyers concur with that assessment, but it is certainly a matter of intense legal discussion and debate.

A final thought for now. The famed British statesman William E. Gladstone once made this memorable remark: "Justice delayed is justice denied." I bring up this quote because some insist the use of public nuisance is perfectly apt and that it is the tool already in the toolkit, ready to go. Waiting for AI laws to be crafted and enacted is going to allow justice to be delayed. Thus, in those eyes, the public nuisance legal tactic is fortuitous, and a welcome expedient means of public relief.