How Courts Are Beginning to Define Liability in the AI Era

These cases aren’t just about money.

They focus on establishing an entirely new legal benchmark for what AI companies owe consumers.

Why The Old Legal Shields Are Failing

For a long time, AI companies hid behind two big legal shields:

1. Section 230 (the rule that protects platforms from user-generated content)
2. The argument that AI output is “speech” rather than a “product”

Both are starting to crack.

Think about it:

A toaster starts a fire that burns down someone’s house. The maker of the toaster gets sued. Why should a chatbot that was allegedly responsible for a teen’s death be any different?

That’s the exact argument plaintiffs are making — and judges are starting to listen.

Here’s why this matters:

Shifting the perspective on AI output from “speech” to “product” fundamentally alters how the courts deal with it. Products can be poorly designed. Products can fail safety inspections. Products can be recalled. And most of all… products can be sued for the damage they cause.

Potential theories of liability in lawsuits like Raine v. OpenAI include negligence, product liability, and wrongful death. Same as if a car had defects. Same as if a medical device misfires.

That framing has huge consequences for the entire AI industry.

How Courts Are Defining AI Liability

How are courts developing the law around AI liability? Three trends.

Foreseeability Is The New Battleground

Courts are examining whether AI firms knew or should have known their products could injure vulnerable consumers.

Here’s the kicker:

OpenAI has publicly acknowledged that safeguards degrade throughout extended conversations. Plaintiffs love it when defendants make admissions like that.

AI companion mental health risk lawsuits are banking on this. A Common Sense Media survey discovered in 2025 that 72% of teens have interacted with AI companions at least once — if not regularly. That means AI companies definitely knew kids were using their services.

Product Liability Is The Leading Theory

The majority of significant AI litigation today is predicated on product liability theory. The complaints have alleged that:

• The product was defectively designed
• Safer alternatives existed
• The company failed to warn users
• The harm was reasonably foreseeable

Consider the case of Soelberg. In December 20 25, a federal judge refused to dismiss claims against OpenAI stemming from a murder-suicide linked to prolonged ChatGPT interaction. Huge development.

Regulators Are Closing The Gap

While courts work through individual cases, regulators are setting broader rules.

In September 2025, the FTC launched a formal inquiry into generative AI developers’ practices around protecting minors from harm caused by chatbots.

Later that same day, bipartisan coalition of 44 state attorneys general issued a joint letter to major AI companies, calling for increased accountability for child safety.

State laws are beginning to follow. In May 2025, New York became first to put regulations in place surrounding AI companions, mandating that providers have systems in place to identify and address any language suggesting suicidal thoughts or self-harm.

California went further with its LEAD Act, which Gov. Newsom vetoed in October 2025. Regardless, the political pressure continues to mount — if anything, it’s building steam.

What This Means For The Future

The legal landscape is shifting faster than anyone expected. Three big changes are coming.

Stronger Product Safety Standards

Expect AI companies to face new product safety standards for sensitive topics like:

• Suicide
• Self-harm
• Mental illness
• Substance abuse

Courts will likely turn voluntary platform measures into legal requirements.

Clearer Duty Of Care

The largest remaining question is how a “duty of care” will be shaped with respect to AI. The contours of this standard are being established on a case-by-case basis presently. In the coming years, courts will develop a more robust understanding of to whom AI companies owe a duty (particularly children), what that duty entails, and how to determine if a company violated that duty.

This will shape AI development for decades.

More Aggressive Enforcement

Congress can hold its breath. State AGs, the FTC, and private plaintiffs aren’t waiting. Legal attacks are coming from all sides — and building steam.

But here’s the important part:

None of this implies AI will disappear. These lawsuits aren’t seeking to eliminate chatbots. They want to make them safer.

Final Thoughts

The legal landscape around AI is being rewritten in real time.

What began as isolated cases has turned into a litigation tsunami that’s altering judicial perceptions of software, speech and safety. Although the law continues to play catch-up with technology, one thing is clear:

• AI companies will be held accountable for foreseeable harm
• Product liability is the leading legal theory in AI cases
• AI companion mental health risks are now a legitimate cause of action
• Regulators are stepping in where Congress hasn’t

For users — and especially parents — the takeaway is simple:

AI chatbots are incredibly useful, but they aren’t therapists, they aren’t friends, and they are not a safety net. Until the courts intervene, the responsible move is to treat AI as what it is: A program with serious limitations.

The era of “AI can do no wrong” is over.

The era of AI accountability is just beginning.

Read More From Techbullion