Two jury verdicts in one week just handed plaintiffs' lawyers a playbook for suing AI companies.
On March 24, a New Mexico jury ordered Meta to pay $375 million after finding the company liable for child exploitation failures on its platforms. The next day, a California jury ruled that Meta and YouTube were negligent in causing a young woman's depression and anxiety from compulsive childhood social media use, awarding $6 million in damages. Meta's stock dropped nearly 8% on the combined news.
The verdicts alone are significant. But the real story is what they mean for the AI industry.
The Research Liability Trap
More than a decade ago, Meta hired social science researchers to study how its platforms affected users. The idea was to show the company took user wellbeing seriously. In court, those same internal studies became evidence that Meta knew its products caused harm and shipped them anyway.
This creates a brutal dilemma for AI companies. Conduct internal safety research, and you generate documents that plaintiffs can subpoena. Skip the research, and you're flying blind on products that interact directly with hundreds of millions of people. A Harvard Law School lecturer described the verdicts as "a major watershed event" representing "a big shift in how Americans are viewing big tech."
AI Companies Already in the Crosshairs
The legal arguments that worked against Meta are already being adapted for AI cases. OpenAI faces more than a dozen wrongful death suits alleging ChatGPT reinforced paranoid delusions. Google and Character.AI face multiple lawsuits involving chatbot-related teen suicides and mental health crises. Character.AI has already settled one case involving minors.
The core claim is identical to what won against Meta: when companies make intentional decisions about how products are built, they bear responsibility for foreseeable consequences.
AI cases may actually be harder to defend than social media cases. Chatbot responses are generated by the company's own models, not posted by users. That distinction could eliminate Section 230 protections - the legal shield that has historically kept tech platforms from being treated as publishers of user content. As Meetali Jain of the Tech Justice Law Project put it, companies must be "held responsible for the foreseeable consequences of those choices."
What Changes Now
Both OpenAI and Character.AI have already added parental controls and safety panels in response to litigation. But the industry remains largely self-regulated, and roughly 2,000 additional social media harm lawsuits are still in the pipeline.
For anyone building AI products, the practical takeaway is stark: internal safety research that was once a PR asset is now a litigation risk. And the legal standard for "defective product design" now clearly applies to algorithmic and AI-driven features, not just physical products. Every design decision - from how a chatbot handles emotional conversations to whether it can refuse dangerous requests - is potential courtroom evidence.