Meta Platforms has reached an $18 billion settlement with a coalition of American states over child safety violations, but the engineer whose evidence helped build the case against the social media giant argues the deal amounts to little more than cosmetic changes that will leave teenagers exposed to the same documented harms. Arturo Bejar, a former safety engineer at Meta, testified in court that the company deliberately underreported how frequently young people suffered harm on its platforms, algorithmically steered them toward content damaging their self-esteem, and failed to adequately investigate child predation complaints. His statements formed a critical pillar of the states' legal action, yet Bejar contends the settlement does not meaningfully confront any of these issues.
California's Attorney General Rob Bonta hailed the settlement as transformative, declaring it would "make social media less dangerous for our kids." Meta, for its part, claimed it had "longstanding efforts to empower parents and support teens." Bejar's assessment, however, cuts against the official narrative. He characterized the settlement as potentially cementing what he calls Meta's "safety theater"—the performance of concern without substantive change. While the agreement might reduce usage slightly, Bejar argued, it will not meaningfully improve the experience for vulnerable young users. This divergence between regulatory celebration and whistleblower skepticism reveals a fundamental tension in how the settlement was structured and what it actually demands of the company.
The remedies included in the settlement often track measures Meta had previously considered and deliberately rejected on the grounds they would have minimal impact on either user wellbeing or business operations. One prominent example involves hiding like and reaction counts from posts, intended to reduce "negative social comparison" among teens. Meta's internal research, known as Project Daisy and dating to 2019, found that obscuring these metrics produced at most modest improvements in user experience. An internal presentation to Mark Zuckerberg in 2020 revealed that Facebook employees observed no meaningful movement in overall wellbeing measures following the change. The business impact was similarly negligible, with tests projecting a decline of just 0.09 percent in daily active users if implemented. Meta ultimately made hiding likes an optional feature for users rather than a default setting, a choice that renders it largely ineffective for those most vulnerable to social comparison harms.
Years of leaked internal documents and legal discovery have documented Meta's understanding of how its platforms damage adolescent mental health. The company's own researchers flagged algorithmic systems that disproportionately recommend fitness and beauty content to teenagers already struggling with self-esteem. Internal studies also revealed intentional design choices engineered to maximize engagement by keeping young users scrolling beyond what they would independently choose. These design decisions were not incidental byproducts but deliberate engineering choices made with full knowledge of their psychological impact. Bejar's work at Meta had specifically concluded that users reported negative experiences on the platforms at rates substantially higher than the company's publicly disclosed figures, suggesting Meta had been systematically downplaying the severity and frequency of harm.
The settlement addresses these documented problems only peripherally. Instead of requiring fundamental changes to algorithmic recommendation systems or forcing transparency about the extent of harms, the agreement focuses primarily on expanding parental controls, implementing time limits on use, and improving detection of age misrepresentation. These remedies treat the symptom rather than the disease. Yet even these relatively modest requirements face enforcement challenges. Meta's age verification depends on user self-reporting, a mechanism demonstrably inadequate to the task. Australia's experience provides instructive evidence: despite the government banning social media use for teenagers under 16, regulators have found that eight in ten young teens remain active on such platforms. This suggests that technical controls alone, without addressing the underlying appeal or addictive properties of the platforms, prove largely ineffective.
Responses from adolescent mental health professionals have been divided, reflecting genuine uncertainty about whether the settlement will produce measurable benefits. Dr. Jane Conron, a clinical psychologist at Northwestern University's Feinberg School of Medicine, expressed skepticism about one specific provision requiring Meta to offer an algorithmically uncurated feed to teenagers on an opt-in basis. She predicted such a feature would be adopted by few users, rendering it meaningless as a protective measure. Opt-in solutions place the burden on the user to select a safer experience, a choice psychologically vulnerable teenagers may struggle to make. However, Conron acknowledged that the settlement's daily time limits on Instagram and Facebook use could prove meaningful for certain young users, particularly those struggling with compulsive usage patterns.
Conron reported that some of her young patients become severely emotionally distressed at the prospect of leaving the platforms, to the point of tears when parents attempt to restrict access. A built-in usage cap administered by the platform itself, rather than requiring parental policing, could provide relief from this struggle. Beyond the direct effects of time limits, Conron suggested the settlement's very existence might catalyze important conversations between teenagers and their parents about social media's role in their lives. The acknowledgment by regulators that these platforms present documented harms, even without Meta admitting fault, may shift the cultural understanding of social media risks among families. The agreement does not require Meta to concede that it built harmful products or accept responsibility for the mental health consequences its researchers documented, a significant limitation that preserves the company's legal position.
The philosophical disagreement between Bejar and Conron illustrates a deeper problem with regulation in the social media era. Bejar emphasizes that superficial changes cannot address architecturally embedded harms—that a platform designed from foundation to maximize engagement while leveraging psychological vulnerabilities cannot be made safe through peripheral adjustments. Conron recognizes the settlement's constraints but perceives potential value in even limited interventions and the precedent of regulatory action. For Southeast Asian observers, this dispute carries particular weight. The region has among the world's highest rates of social media adoption among teenagers, with correspondingly high rates of reported cyberbullying, self-harm, and mental health crises linked to platform use. Malaysian policymakers and parents watching the Meta settlement will need to determine whether similar approaches would adequately protect young people in this region's context.
The structural weakness of settlements like Meta's becomes apparent when examining what they fail to require. The company will not be forced to disclose how frequently its algorithms promote harmful content to vulnerable users. It will not have to redesign its fundamental business model, which depends on maximizing time spent on platform and the psychological engagement that drives advertising revenue. It will not have to explain why internal researchers' warnings about youth mental health were disregarded in favor of engagement metrics. Instead, Meta pays a substantial but ultimately manageable sum—spread across a decade and offset by the company's vast revenues—and implements a suite of parental controls and usage limits that place responsibility on families rather than fundamentally altering what Meta offers.
For Malaysian stakeholders, the settlement's limitations suggest that waiting for international regulatory resolution is insufficient. The country's own policymakers and industry bodies must grapple with whether the protections offered by Meta to American teenagers would adequately serve Malaysian youth, who may face additional vulnerabilities including language-specific harassment, cultural pressures around social status, and inadequate local mental health resources. Parents and educators in Malaysia watching this case might reasonably conclude that personal and institutional vigilance remains necessary regardless of what enforcement actions multinational companies face overseas. The Meta settlement, while superficially impressive in dollar terms, ultimately demonstrates that regulatory settlements can be negotiated in ways that preserve a company's core business model while creating the appearance of meaningful change.
Bejar's continued insistence that the settlement codifies safety theater rather than genuine protection speaks to a broader reckoning unfolding in technology regulation worldwide. Whistleblowers who expose internal evidence of harm sometimes find that legal victory produces hollow remedies because settlements are negotiated documents that both parties can frame as acceptable. Meta can describe the agreement as demonstrating its commitment to safety. State attorneys general can declare victory before voters concerned about their children's wellbeing. Bejar can speak truth about the settlement's inadequacy, yet the company will still implement precisely what regulators negotiated. Meanwhile, the fundamental question—whether social media platforms designed to maximize engagement can ever be truly safe for adolescents—remains unresolved by a settlement that never forces that conversation into public view.
