AI Product Liability: The Light-Touch Law with Heavyweight Impact

A proven approach to making AI safer, fairer, and more trustworthy for the public.

Shutterstock: 2628542781

Evidence is mounting that AI products — from general-purpose chatbots to so-called “AI companions” — are already inflicting real harms on Americans.

Behind each headline are stories of AI design patterns that manipulate users, induce emotional distress, and shatter trust. Once people grasp the scale of these harms, a natural question follows: What can be done?

One of the most straightforward and effective solutions lies in product liability — a tried-and-tested legal approach that motivates safer product development, holds companies accountable when their products cause harm, and is light-touch enough to support American innovation.

Applied to AI, it would be a powerful way to turn the tide on these harms.

What Is Product Liability?

In simplest terms, product liability holds companies and manufacturers legally liable — or responsible — for harms their products cause. The history of this legal approach stretches back to the 19th century. Today, product liability is the norm for consumer and business products. From the cars you drive, to the foods you eat, to the medicine in your bathroom cabinet, product liability is what ensures American products are reliable, trustworthy, and safe.

---

---

How Does Product Liability Work?

Practically speaking, product liability impacts two streams:

1. Preventative stream, and

2. Responsive stream

Preventative: When a company knows it could be held liable for harms its product causes — which product liability establishes — the company is far more likely to prioritize safety in its product development process. To use automobile companies as an example, that includes designing vehicles with consumer safety front-of-mind (think seatbelts, airbags, antilock brakes, etc.), carrying out robust crash testing, and more. If a company could be held liable for harms, they’re more likely to ask, “How can we prevent harms from happening?”

Responsive: If harms do occur once a product is out in the world, product liability gives consumers and businesses clear legal pathways to hold companies responsible in court. This brings a sense of clarity to consumers that accountability is within reach if a defectively designed product harms them.

It’s important to note that product liability does not tell companies or industries exactly how to design their products. It simply requires companies to prioritize safety during their development and manufacturing processes. Because of this, the approach is considered “light touch” and innovation-friendly.

Does Product Liability Apply to AI?

Not yet. But there are signs that the legal system has already begun viewing AI as a product, given in recent court decisions. And deeming AI a “product” opens the door to a product liability approach.

Generally speaking, the digital tech industry has fought to remain the exception to product liability in America — with social media and AI companies often fighting the hardest. In previous decades, courts deemed software a “service” instead of a “product,” but it was primarily the tech companies themselves who pushed to uphold this framing in the years that followed. In the 2010s and 2020s, social media and AI companies seized on this line of legal thinking, since “services” are not held to the same legal standard when it comes to responsibility to the consumer. By pushing to maintain this “service, not product” legal framework, today’s digital tech companies have further avoided accountability for the harms their platforms cause.

But perspectives are evolving. Legal teams are increasingly challenging the tech industry’s position and arguing that these platforms should be classified as “products” subject to liability standards. In the court of public opinion, technologists, including CHT’s co-founders, have demonstrated just how much tech platforms are being designed, manufactured, and sold to consumers — all of the hallmarks of a “product.”

As a result, AI is now increasingly considered a “product” by the courts and public. This includes popular platforms such as ChatGPT, Character.AI, Claude, Gemini, and more. When AI is labeled a “product,” it can be regulated with a product liability approach.

What Happens Next?

Lawmakers and advocates — including Center for Humane Technology — are actively championing a product liability approach to AI in policy spaces. This would hold artificial intelligence to the same legal standards that other trusted American products are held to.

The AI LEAD Act, introduced by Senator Dick Durbin and Senator Josh Hawley in September 2025, shows that momentum is building around federal product liability — and, crucially, that this approach has bipartisan support. There have also been several liability bills introduced at the state-level this past session. We can anticipate more to come in 2026.

These legislative processes are in the early stages. But what they demonstrate is clear political and judicial will to apply product liability to AI.

How Could AI Product Liability Improve Society?

If product liability were applied to AI products, Americans would be assured that the AI products available for download in app stores, readily available in their browser, or put into the stream of commerce have been designed with safety in mind from the outset. This would include the most popular chatbots on the market today.

So, what would that look like in day-to-day life?

While we can’t predict exactly how these AI products would be designed (again, the approach does not prescribe design changes), we can look to other industries that have product liability applied, and imagine AI products with safety features programmed as the default option. We can imagine clear, accessible, and publicly available safety reporting on AI products, reporting that brings clarity to the risks of use. We can imagine AI products with clear warning labels for mental health, and resources for seeking human support. We can imagine AI products that are quicker to end dangerous conversations, including conversations related to self-harm and suicide. When it comes to businesses using AI, we can imagine AI business products with clear warnings about potential product failures, which would empower business owners with more information about the technology they’re integrating into their workstreams.

Just as product liability helped incentivize safer car designs, safer food manufacturing, and safer medicines, it can incentivize safer AI products for all of society to use. It’s time to update our laws for the 21st century, and apply product liability to AI.