A Whistleblower Just Told a Court Meta Hid the Scale of Harm to Teenagers

Key Takeaways

• The trial of Meta — brought by 29 US states — opened in a federal court in California on August 18th, with states alleging Meta deliberately designed its platforms to be addictive to young people and hid internal research showing the scale of harm.
• Meta whistleblower Arturo Béjar testified that Meta's own user surveys showed teens were exposed to graphic or violent content at a rate 100 to 400 times higher than the 0.01–0.02% Meta reported publicly — and that Meta chose not to publish those survey results.
• Béjar, a former safety executive who worked with Zuckerberg directly, said safety and security were "an afterthought" during the company's "move fast and break things" era, and that harm findings were detected but changes weren't made.
• The trial is expected to last five to seven weeks. Cornell law professor James Grimmelmann said a Meta loss "could have ripple effects for a lot of platforms" and fundamentally change how social media companies can be held liable for interface and design decisions.

Day One: What the Whistleblower Said

The biggest trial Meta has ever faced opened in a federal court in California on August 18th, brought by four states acting on behalf of 29 that filed a lawsuit in 2023 alleging Meta violated child privacy laws and deliberately designed its platforms to addict young users, according to BBC reporting. 

The states are seeking significant financial penalties; California Attorney General Rob Bonta described Meta's conduct as "illegal" and accused the company of choosing profits over safety.

The first witness was Arturo Béjar, a former Meta safety executive who worked closely with Mark Zuckerberg. His testimony was direct and specific. Meta, he said, reported that the risk of teens being exposed to graphic or violent content was 0.01 to 0.02%. 

Meta also had user survey results it didn’t publish showing that the prevalence of graphic or violent content was 100 to 400 times higher than that figure. When asked to define how Meta measured harm, he was equally direct: prevalence of policy-violating content "is a very narrow measure of a very specific thing." Harm, he said, is "what people experience as harm."

Béjar returned to Meta in 2019 partly because of his own daughter's experience. Shortly after setting up an Instagram account at 14, she began receiving unsolicited sexual messages from strangers. When he asked if she'd reported them, she told him there was no option to do so. "I just block them and they laugh it off and they just move on to the next person," she told him. He thought something had broken in the company and wanted to fix it from the inside.

He described the period from 2019 to 2021 (when Meta's internal mantra was "move fast and break things") as one in which safety and security was "an afterthought." Research detected significant harms, including to young people, but changes were not made based on those findings. 

He emailed Zuckerberg directly after a 2021 post in which the CEO said Meta didn’t prioritize profit over user safety, telling him the post created a "false and misleading impression" of the company's actual commitment.

What Both Sides Are Arguing

California's deputy attorney general Megan O'Neill framed the state's case as a split screen: public statements from Meta executives about their commitment to safety, set against what internal research actually showed. That framing — that Meta knew, didn't act, and then misrepresented the situation to the public — is what the prosecution will spend the next five to seven weeks trying to establish.

Meta's lawyer Paul Schmidt argued the company has a track record of transparency, has used internal research to develop tools that protect young users, and that some internal documents will contain language that is "loose" — a pre-emptive effort to blunt the impact of emails and messages that jurors will see in the coming weeks. 

Meta "strongly disagrees" with the allegations, the company said in a statement, and evidence will show its "longstanding commitment to supporting young people."

Cornell law professor James Grimmelmann told the BBC the case could have consequences well beyond Meta: "The core idea that you could be held liable for interface and design decisions... that cuts across the entire set" of social media platforms. A loss, he said, would likely force Meta to remove features "most appealing" to minors and could accelerate legislation limiting young people's access to social media more broadly.

Why This Trial Matters for the Bigger Picture

We've written about the age verification and social media ban landscape — from Australia's ban built partly on a report with AI-hallucinated citations, to France's constitutional court striking down its under-15 ban as disproportionate, to the KIDS Act and SCREEN Act advancing in the US Senate. The political argument driving all of that legislation is that platforms are harming young people and aren't doing enough about it.

The Meta trial is where that argument gets tested in a courtroom with evidence. If Béjar's testimony holds — that Meta had internal data showing harm at 100 times the publicly reported rate and chose not to act on or disclose it — that's a materially different situation from "platforms are trying but struggling." That's a company that knew, didn't tell, and kept building the same way.

That distinction matters for how we think about the legislative response. Age verification mandates and platform bans are a particular kind of answer to the problem: they restrict access. Liability for design decisions is a different kind of answer: it changes the incentive structure for how platforms are built. 

The states in this courtroom are arguing for the second approach — hold Meta accountable for what it built and what it knew, rather than building surveillance infrastructure around the edges of it to manage the consequences.

That argument, if it succeeds, is more targeted and more proportionate than anything currently working its way through legislatures. Whether it does succeed — and whether Meta can mount a credible defense against testimony from its own former safety executive — is what the next five to seven weeks are for.