As artificial intelligence tools become embedded into daily life, litigators are beginning to see a shift away from social media disputes and toward a new wave of claims testing how courts assign liability when technology itself is constantly evolving, said Liz Och, a Denver-based attorney with Hogan Lovells.
“AI litigation generally covers almost every base and continues to find new bases every day,” she said, pointing to claims spanning employment, antitrust and beyond. But the cases drawing the most attention, she said, are tort-based claims seeking to hold AI companies accountable for personal injury and similar harms.
Those claims echo earlier litigation against social media platforms, though Och said the comparison is limited. “It comes down to what I would think of as active versus passive,” she said.
In social media litigation, disputes often center on content hosted on a platform: What was posted, what was removed and what companies allegedly allowed to remain? AI systems, by contrast, generate content in real time through user interaction.
“Interacting with a robot that has been trained to respond and to act in certain ways” introduces a more active dynamic, with systems “creating content as you go,” she said.
That distinction is surfacing in early lawsuits, including claims involving AI chatbots. Some cases, Och noted, allege AI systems contributed to serious harm, especially involving minors. While several cases filed in Colorado have been resolved quickly, she expects “it’s a matter of time” before these distinctions are explored in court.
A Moving Target for Liability
The generative and evolving nature of AI systems complicates one of the core concepts underlying tort law: foreseeability.
As courts confront AI claims, they will be asked to evaluate what risks were foreseeable at the time decisions were made — a moving target in a rapidly developing technology landscape.
Och pointed to a now well-known example for lawyers: AI systems generating inaccurate or fabricated legal citations. “That’s a very common risk” today, she said. “But that’s not really something that was known a year and a half ago.”
A future knowledge base raises difficult questions for courts and juries tasked with looking backward. “It could be that a jury is sitting there in 2030, and they’re looking back to today,” Och said. “Let’s try to put ourselves in 2026 and think about what was foreseeable.”
Old Doctrines, New Applications
Without comprehensive legislation governing AI, plaintiffs are relying on existing legal frameworks, particularly negligence, to bring claims.
“They’re having to kind of pick and choose and figure out what existing contexts they can fit their claims into,” she said.
While negligence provides a familiar structure (duty, breach, foreseeability and damages), applying negligence concepts to AI systems has no clear precedent. “The challenge becomes, without any precedent, courts are going to have to try to find something analogous,” she said.
The process could lead to divergent outcomes across jurisdictions, depending in part on how comfortable judges are with understanding the technology and the range of actors involved in building and deploying it.
Who Holds Responsibility?
Another unresolved issue is where liability will land.
While early cases have targeted AI developers and platform providers, other actors include companies modifying models and businesses deploying AI tools in their operations.
As more AI cases reach discovery, plaintiffs may begin identifying additional potential defendants and refining their theories of liability.
“All companies need to be really mindful (when using AI) and need to make sure they’re either mitigating the risk or at least keeping good records,” Och said, “so that they have a good story on the back end if they are faced with a lawsuit.”
For now, “every new claim that is filed is another kind of interesting look at where things may go.”