Snap settled a lawsuit this week that claimed its technology caused severe psychological distress to two teenage users, resolving the case confidentially and sparing chief executive Evan Spiegel from testifying in a Los Angeles court, CNBC reported. Meta, TikTok and YouTube had already settled with the plaintiffs in the same suit, one of several tied to a social media personal injury trial that Meta and YouTube lost earlier this year. More trials are scheduled to begin in March.

The same week brought a new complaint against Google, filed by the parents of a child who they say suffered a psychological crisis after chatting with the Gemini AI assistant inside Google Classroom. The plaintiffs argue that Google integrated Gemini into school Chromebooks despite known risks. Google said it is reviewing the claims and pointed to features built with medical and academic advisers.

The two events belong to the same story, and it is now a financial one. Litigation that began as a design complaint against social platforms has become a disclosed risk in artificial intelligence offerings, including Anthropic's prospectus for a listing this month. Michele Neitz, a professor at the University of San Francisco School of Law, told CNBC the disclosure is a clear sign that companies expect more litigation than they did a decade ago.

Why the cases target design

Section 230, passed in 1996, shields platforms from suits over content their users post. Plaintiffs' lawyers and state attorneys general have therefore built cases around product design instead: the mechanics of engagement, recommendations and notifications, which the platform controls.

That theory survived a jury. Meta agreed in August to pay nearly $17 billion to settle with a group of state attorneys general in a child mental health case, and some of those offices are pursuing similar claims against its rivals. A federal case involving Meta, TikTok, Snap and YouTube alongside a long list of school districts is still running.

Snap's settlement, according to Joseph McNally, director of emerging litigation at McNicholas & McNicholas, is part of an effort by tech companies to turn a corner. "Not only because I think that there were real legal challenges, but also the reputation of their products, the reputation of their companies has really been hit," he said. A Snap spokesperson said the parties resolved the matter amicably and described safety work as never finished.

AI moved the argument into classrooms

The Google case matters because the product is in a school, distributed by an institution rather than chosen by a child. It also lands next to a distribution fight: the New York Times reported that K-12 schools which block Google's AI tools for students would lose access to Search, which is the subject of Google's earlier standoff with districts that refused its AI answers. Chris Phillips, Google's vice president for education, told the Times that "Search and AI are deeply integrated."

Kate Blocker, director of research and programs at the nonprofit Children and Screens, said the personalized nature of chat products makes it hard to predict what they will say to a child, and argued that developmental safety has to be tested before release rather than after harm appears. "I think it's so so important that children's safety is being considered when these products, AI, social media, et cetera, are being designed, and not after the fact as these problems emerge," she said.

The cost is arriving before the revenue

For investors, the relevant question is how these claims price into companies that are still losing money. OpenAI faces suits from state attorneys general and private plaintiffs that borrow the design theory from the social media cases. Anthropic, which prohibits children from using Claude, still put the exposure in its filing, a decision that reads as an acknowledgement that the risk attaches to the category rather than the product.

The pattern from social media is that settlements arrive one at a time and the totals are disclosed late. Meta's $17 billion figure came years after the first complaints. If the AI cases follow the same path, the industry's legal costs will be set by trials scheduled for next spring, and by whether courts accept that a chatbot's design, like a feed's ranking, is the company's own act.

If you are in distress or having suicidal thoughts, the Suicide & Crisis Lifeline can be reached at 988 in the United States.