Arturo Béjar spent eight years inside Meta, spoke to Mark Zuckerberg at least 100 times, and ran the surveys that produced the damning numbers. Fifty-one percent of teen Instagram users reported bad or harmful experiences in the previous seven days. Of those, content was removed 0.02% of the time. Béjar sent these findings directly to Zuckerberg and Instagram CEO Adam Mosseri. Neither acted on them. The structural logic is not complicated. Béjar testified that infinite scroll — the feature that keeps users swiping — is the revenue engine: "the more views you have, the more ads you sell, the more revenue you make." Fixing child safety would mean interrupting that loop. Removing predatory content recommendations, limiting engagement-maximizing algorithms for minors, building functional reporting tools — all of these reduce time-on-platform, which reduces ad impressions, which reduces revenue. The business model and child safety are in direct tension, and Meta chose the business model. The trial, brought by 29 state attorneys general in Oakland federal court, alleges Meta deliberately designed addictive products that harmed minors and illegally collected data on children under 13. The potential liability is staggering: up to $200 billion, equivalent to Meta's entire 2025 annual revenue. The states are also seeking forced design changes — a remedy that would structurally alter Meta's engagement-driven architecture. Béjar's testimony carries particular weight because he is not a disgruntled outsider. Meta's own lawyers established this on cross-examination, confirming he had Zuckerberg's support, left on good terms, and remains proud of his work at the company. The implication Meta wanted to plant — that Béjar is a loyalist, not a critic — cuts the other way entirely. A loyalist who sends documented harm reports to the CEO and receives no response is not evidence of a functioning safety culture. It is evidence of a CEO who chose not to read the mail. The personal dimension sharpens the institutional one. Béjar first engaged with the problem after his own teenage daughter received unwanted sexual advances, genital images, and misogynistic abuse on Instagram — and discovered that the platform's reporting tools were either broken or absent. He then scaled that personal observation into systematic research, producing the 51% harm figure. Meta had the data at the individual, survey, and executive-briefing levels. The information asymmetry was not between Meta and the public; it was between what Meta knew and what Meta chose to do. Meta's defense, delivered by attorney Paul Schmidt, rests on two claims: the company built safety tools, and it disabled more than one million under-13 accounts. Neither addresses Béjar's core testimony — that harmful content was removed 0.02% of the time and that the CEO was personally informed and did nothing. Building tools that don't work and removing accounts after the damage is done is not a safety program. It is a compliance theater. The trial is expected to last at least six weeks, with Zuckerberg and Mosseri both scheduled to testify. The internal documents and emails already entered as evidence will be as important as the live testimony. What matters now is whether the jury sees Meta's conduct as negligence — a failure to act — or as a deliberate architectural choice to preserve the revenue loop at the expense of children. Béjar's testimony points firmly toward the latter.
