Arturo Bejar, an ex-Meta safety engineer whose testimony formed a cornerstone of a child protection lawsuit against Meta Platforms, has challenged the adequacy of a settlement announced this week between the company and a coalition of U.S. state attorneys general. Speaking after the deal was disclosed, Bejar contended that the agreement fails to address the substantive problems he detailed during his court appearance, raising questions about whether the remedies will deliver genuine protective benefits to young users.
The settlement, brokered with California Attorney General Rob Bonta and his counterparts across multiple states, will see Meta commit up to $18 billion over the coming decade alongside a series of operational changes designed to curb potential harms. Bonta framed the agreement as transformative, asserting it would make social media demonstrably safer for children. Yet Bejar's critique suggests the measures represent incremental adjustments rather than systemic reforms addressing the root causes of youth vulnerability on Meta's platforms.
During his earlier testimony, Bejar presented evidence that Meta had systematically underestimated the harm teenagers experienced on its services, actively steered users toward content designed to trigger negative self-perception, and demonstrated inadequate procedures for addressing child exploitation. These allegations formed the evidentiary backbone of the states' legal case. However, Bejar now characterises the settlement as essentially formalising what he terms Meta's "safety theater"—a pattern of superficial interventions that create the appearance of change without substantively reducing platform dangers. His stark assessment: Instagram may see slightly lower usage volumes, but the environment itself remains fundamentally unsafe.
Meta's official response emphasised its commitment to teen welfare, positioning the settlement as an extension of existing protective measures. The company pledged to empower parents and bolster support systems for adolescent users. Bonta acknowledged in his statement that broader legislative and industry action remains necessary, yet defended the settlement as introducing tangible, enforceable protections across the nation's youth demographic. This disagreement between the architect of the case and the company highlights the persistent tension in regulating platforms whose business models rely on engagement metrics.
The remedial framework includes measures that internal Meta documents—subsequently leaked—reveal the company had previously considered and deliberately rejected. A prime example involves hiding like and reaction counts on posts, a mechanism intended to mitigate what researchers term "negative social comparison" among young users. Meta initiated testing on this concept through an initiative called "Project Daisy" in 2019, stemming from internal research flagging social media's potential mental health toll. Testing concluded that concealing like counts produced at most modest improvements in user wellbeing, with internal communications from 2020 explicitly stating that Facebook engineers observed no meaningful shifts in overall wellbeing indicators following the intervention.
When Meta assessed the business implications, the calculus proved similarly unpersuasive: modelling suggested daily active users would decline by merely 0.09 percent if the feature became mandatory. Consequently, the company opted to permit users to voluntarily hide likes rather than implementing the change universally. The fact that this same measure now forms part of the settlement illustrates Bejar's contention that Meta is deploying previously discarded interventions, relabeled as concessions. This pattern raises fundamental questions about whether settlements negotiated between state regulators and corporate defendants can effectively challenge business models fundamentally centred on maximising user engagement.
MetaInternal research conducted by Meta's own teams has documented algorithmic systems that disproportionately recommend fitness and beauty content to adolescents already grappling with self-esteem difficulties, alongside design features engineered to extend user session duration beyond what individuals would independently select. The handling of user reports and content moderation emerged as another critical concern, with Bejar's own work establishing that harm incidence exceeded Meta's publicly reported prevalence figures. These allegations occupied central ground in the states' litigation strategy yet receive only peripheral attention in the settlement document. Instead, the accord concentrates on expanded parental oversight mechanisms, session time restrictions, and enhanced systems for identifying minors misrepresenting their age as adults.
The practical effectiveness of age-verification mechanisms deserves particular scrutiny. Meta's capacity to identify and remove underage users depends substantially on self-reported age information, a methodology riddled with gaps. Australia's experience proves instructive: despite implementing a legislative ban on social media access for users under sixteen, regulators discovered that approximately eight in ten young teenagers maintain active accounts on platforms. This enforcement challenge suggests that procedural remedies addressing account demographics operate at the periphery of the core problem.
Responses from youth mental health professionals have proven decidedly mixed. Dr. Jane Conron, a clinical psychologist at Northwestern University's Feinberg School of Medicine, expressed scepticism about provisions requiring Meta to offer non-algorithmically curated feeds on an opt-in basis, predicting minimal uptake given user preference for algorithmic recommendations. However, Conron acknowledged that daily usage caps imposed on Instagram and Facebook access could yield meaningful outcomes for certain young users. She cited clinical observations of adolescents experiencing acute emotional distress when attempting to disconnect, with some patients displaying such profound attachment that parental restrictions trigger crying episodes. Built-in usage limits could provide structural relief, she suggested, potentially reshaping household discussions surrounding digital consumption.
Conron further noted that while the settlement stopped short of requiring Meta to admit fault or accept allegations of intentional product design flaws, its very existence might catalyse broader awareness of platform-related harms. The regulatory action and public scrutiny, even without explicit liability admission, could shift social consciousness regarding youth safety online. She characterised this consciousness shift as potentially valuable for reshaping the cultural conversation between parents and teenagers about technology consumption patterns. Yet this optimistic interpretation coexists uneasily with Bejar's warning that the remedy framework represents regulatory capture dressed in protective language.
For Malaysian and Southeast Asian observers, the settlement carries particular significance. Regional governments face mounting pressure to regulate tech platforms while lacking the regulatory sophistication and enforcement resources of U.S. state governments. If Meta has successfully negotiated an $18 billion settlement that Bejar argues leaves core harms unaddressed, this template may influence how regional regulators approach platform accountability. The settlement demonstrates both the potential power of coordinated state action and its inherent limitations when negotiating with companies whose engineering capabilities and financial resources dwarf most governmental capacity. The question haunting regulators across the region: if American attorneys general cannot secure systemic change, what realistic expectations should smaller jurisdictions maintain?
The Meta settlement ultimately illustrates a recurring regulatory paradox: settlements often deliver financial compensation and procedural modifications while preserving fundamental business model architecture. Meta's algorithms remain optimised for engagement; its recommendation systems continue prioritising content likely to elicit emotional response; its data infrastructure maintains the same underlying incentive structures. The settlement represents compromise rather than transformation—a recognition of reputational and legal risk managed through financial and marginal operational concessions. Whether such arrangements constitute meaningful child protection or sophisticated risk management remains the central question, with Bejar's testimony serving as a counterweight to official celebrations of regulatory victory.
