On August 19, 2026, former Meta safety engineer Arturo Béjar testified in federal court in Oakland, California. The civil suit, brought jointly by attorneys general from 29 states, just opened and is expected to run at least six weeks. It alleges Meta deliberately designed addictive products that harm young people and collected data on users under 13 without parental consent, seeking damages of up to $200 billion — roughly Meta’s annual revenue — plus court-ordered changes to how the products are designed. Béjar’s testimony is the most closely watched moment of the trial so far.
The Witness: An Engineer Who Briefed Zuckerberg a Hundred Times
Béjar spent roughly eight years at Meta across two stints and also consulted on Instagram’s safety work; he left in 2021. He is no anonymous critic: since departing, he has testified before a US Senate committee and appeared in other child-harm cases, building a public record around exactly the claims at issue here. Reuters’ account of the same testimony underscores how unusual it is for a company’s own former safety engineer to anchor a state-led case of this size. He estimated he briefed Zuckerberg on product issues at least a hundred times, and in 2021 sent an email warning that reports of harm to teen wellbeing kept rising. His reason for escalating upward rather than fixing things quietly, he said, was simple: “when Mark makes something a priority, mountains move.” Those reports never got a reply.
‘Don’t Ask, Don’t Tell’ — and Two Numbers
Béjar described Meta’s internal posture on teen safety as “don’t ask, don’t tell”: the company knew its products caused harm, including recommendation systems pushing sexual-predator content and graphic imagery to teenagers. His numbers were the most damning part — internal surveys found 51 percent of teen users had experienced something harmful in the previous seven days, and only 0.02 percent of that content was removed. The gap between those two figures is the core of his argument: measurement existed inside the company, and removal did not follow. If the 0.02 percent figure holds up under evidence, it quantifies the distance between detecting harm and acting on it. He also had a personal reason for testifying: his own daughter received unwanted sexual advances and misogynistic insults on Instagram. Zuckerberg, he said, “created a false and misleading impression of Facebook’s commitment to young people.” A hundred briefings is not a stray complaint; it is a paper trail.
The Courtroom Fight: Meta’s Defense
Meta denies all the allegations. Its attorney, Paul Schmidt, told the court there is “no dispute” that people can struggle with social media, but argued Meta has built tools to address the problem, bars under-13 registration, and has disabled more than 1 million accounts belonging to young users. Cross-examination took a different line, focusing on Béjar’s relationship with the company: he confirmed he had Zuckerberg’s support, left on good terms, and remains proud of the work he did there. In other words, Meta did not dispute his experiences or his access — it disputed his conclusions, betting that a well-treated insider is a less sympathetic whistleblower.
Why This Trial Matters
This is one of the few lawsuits with a real chance of changing product design itself: beyond money, the states are asking the court to force changes to how Meta’s products work, which would set a precedent for every platform that measures teen engagement. Zuckerberg and Instagram chief Adam Mosseri are both on the witness list, along with outside experts, so the coming weeks will show whether the company’s account of its internal warnings matches the documents. The procedure is worth watching too: discovery will pull years of internal surveys, emails, and product reviews into the public record, much of it material Meta never intended to publish. The states’ theory — that harm was measured, reported, and deprioritized — lives or dies on those documents. For platform engineering and product teams, the trial puts one question in the open: when internal data shows the product is harming users, what happens after the report goes up the chain? Béjar’s answer — a hundred briefings and one unanswered email — may become the standard citation in every future debate about platform governance.
Sources
AI-assisted summary compiled from the sources above, reviewed by a human before publishing.