Meta's first witness told a jury the blocker on teen safety was culture, not code

Arturo Bejar, the first witness the four state attorneys general called, told a federal jury in Oakland on Wednesday that the reason Meta could never fix its own safety problems was the culture Mark Zuckerberg had built ^[1]. That framing shifts the case away from whether harm is provable in the abstract and toward something narrower and more discoverable: how a product organisation decides what ships, and what it leaves as an option.

The trial opened Tuesday before Judge Yvonne Gonzalez Rogers in the Northern District of California, with California, Colorado, Kentucky and New Jersey arguing that Facebook and Instagram were engineered to hook minors and keep them scrolling ^[2]. The states compressed the business model into three verbs in their openings: hook, hold, harvest ^[3]. Bejar worked inside Meta for roughly eight years on these problems, from 2009 to 2015 and again from 2019 to 2021, and the second stint ended with an internal warning about teenage harm that he later took to the US Senate ^[4]. In court he put it this way: "At the end of the day, it was the company culture that Mark had created that made it so that it was practically impossible to deliver features that addressed the wellbeing and safety issues" ^[5]. Asked about a 2021 Zuckerberg post denying that Meta put profit ahead of safety, he said it did not match what he had seen ^[6].

The operationally useful part of his testimony is not the culture line but the mechanism. The states' design case rests on infinite scrolling, autoplaying video, beauty filters and the like button, and Bejar's contribution was that the countermeasures Meta did build were shipped as optional settings rather than defaults, which in practice meant almost nobody used them ^[7]. Anyone who has run an experiment knows what a default is worth. It also means the interesting artefacts are internal: the escalation that was written, the review where the default was rejected, the document the states are now quoting.

Meta's cross-examination leaned the other way. Brian Stekloff pressed Bejar on whether he or his teams had actually solved any of the harms he described, aimed less at his credibility than at the premise that these problems have buildable solutions Meta declined to build ^[8]. In openings, Meta counsel Paul Schmidt said the company had been alert to teen risk, had worked on it, and had not misled the public; Meta's broader position is that the states assembled the case from selectively quoted internal documents ^[9].

The stakes are contested and the figures do not agree. Meta has told the court the states are pursuing penalties that could reach $1.4 trillion, a number the states have not adopted, while California Attorney General Rob Bonta has framed exposure against roughly $200bn in annual revenue ^[10]. On Meta's own arithmetic, that is about seven years of revenue ^[16]. A third figure surfaced when Gonzalez Rogers ruled Bejar could testify over Meta's objection, reported alongside an estimate that the states' claims might reach $200bn ^[13].

Oakland is not the only front: a New Mexico court ordered Meta to pay $567m into a teen mental health fund this year, and a Tennessee jury is weighing Instagram's design in a seven-week trial ^[11]. Because the states sue under consumer protection and children's privacy statutes, the bar for proving harm to any individual user is lower than in a personal case ^[12].

Watch whether the default-off account holds up as the bridge the states need between harm and knowledge, since they must show both that the products hurt young users and that Meta knew and said otherwise ^[14]. The trial is set to run about six weeks, and the states are still declining to say whether they will call Zuckerberg ^[15].