Whistleblower Testimony Reveals Meta Ignored Teen Harm Data for Years

A former Meta safety engineer told a California jury this week that he warned Mark Zuckerberg directly, and repeatedly, about predators and harmful content reaching children on Facebook and Instagram. According to Meta whistleblower testimony, the CEO never wrote back.

Arturo Béjar spent two days on the stand as the first witness in a trial that could reshape how the world’s largest social media company handles young users. He said he briefed Zuckerberg on product issues roughly 100 times during his career at Meta, and that one 2021 email specifically flagged constant reports of harmful content and damage to teen wellbeing.

What Béjar Told the Jury

Béjar’s account centers on a gap between what Meta said publicly and what it knew internally. He testified that the company was aware its recommendation systems were pushing content from sexual predators alongside violent and graphic images to young users, and that he raised the problem with Facebook and Instagram executives without seeing much change.

The email to Zuckerberg came after the CEO stated publicly that Meta doesn’t prioritize profit over safety. Béjar said that claim didn’t match what he was seeing inside the company. “I felt that he created a false and misleading impression of Facebook’s commitment to young people,” he testified.

He said he escalated issues straight to the top because of how Meta actually worked. “In my experience, when Mark makes something a priority, mountains move,” Béjar told the court. When asked if Zuckerberg ever responded to the 2021 email, his answer was one word: no.

The Numbers Behind the Testimony

Béjar’s most concrete evidence came from surveys he ran on teen experiences on Instagram. He found that 51% of teenage users reported having a bad or harmful experience on the platform within the previous seven days. Of the content those users flagged, only 0.02% was ever taken down.

His interest in the issue wasn’t purely professional. He testified that his own teenage daughter received unwanted sexual advances and explicit images on Instagram, along with misogynistic insults, and that Instagram’s official reporting tools were either useless or unavailable when she tried to use them. That experience is what pushed him to start surveying other teens in the first place.

Meta’s attorney, Paul Schmidt, tried a different approach in cross-examination, not disputing the numbers directly but establishing that Béjar left the company on good terms and still respects the people he worked with. Béjar agreed with each question. It’s a strategy that doesn’t contest the data so much as it tries to make the witness look less like an adversary and more like someone airing a disagreement, which may or may not land with a jury that’s already heard the 0.02% figure.

The Case Itself

Twenty-nine state attorneys general brought the suit, arguing Meta knowingly built addictive products aimed at young users and separately collected data on children under 13 without parental consent, a violation of both federal and state law. California deputy attorney general Megan O’Neill put it plainly in her opening statement: “Meta didn’t do its share.”

Meta denies all of it. Schmidt acknowledged in his opening that people can struggle with social media but argued Meta has “come up with tools to try and address that,” pointing to the fact that Meta doesn’t allow under-13 registration and has disabled more than a million underage accounts.

The trial is set to run at least six weeks in Oakland, with Zuckerberg and Instagram CEO Adam Mosseri both expected to testify alongside company executives and mental health experts. If the states win, Meta could face damages as high as $200 billion, roughly matching the company’s 2025 annual revenue, along with court-ordered changes to how its products are designed.

A Familiar Gap Between What Companies Know and What They Fix

The pattern Béjar describes isn’t new to tech accountability cases, but the specificity is what sets this one apart. Most whistleblower complaints against social media firms rely on leaked documents or anonymous sourcing. This one comes from someone who briefed the CEO in person roughly 100 times and still, by his own account, left on good terms. That’s a harder narrative for a defense team to dismiss as a disgruntled ex-employee settling scores.

The 0.02% removal rate against a 51% harm-reporting rate is the number that will likely define this trial regardless of how the legal arguments unfold. It quantifies something critics of social media have long argued but rarely could prove with a company’s own internal data: that reporting tools exist more as a compliance checkbox than a functioning safety mechanism. If Meta’s own research produced that gap and executives saw it, the “we didn’t know” defense becomes much harder to sustain than in previous cases like Frances Haugen’s 2021 disclosures, which dealt more in policy documents than in a live paper trail running to the CEO’s inbox.

The business-model argument matters just as much as the safety failures. Béjar’s testimony ties infinite scroll and engagement-maximizing design directly to ad revenue, which reframes the case from a content-moderation dispute into a question about whether the product itself is built to work against user well-being by design. That’s the argument that could produce the court-ordered redesign the states are asking for, and it’s also the argument every other platform running similar engagement mechanics will be watching most closely, since a verdict against Meta on those grounds sets a precedent that doesn’t stay contained to one company.