A former Meta safety engineer at the heart of a high-profile child protection case has publicly questioned whether the social media company's $18 billion settlement with a coalition of American states will genuinely protect teenagers from documented harms. Arturo Bejar, who testified about Meta's knowledge of dangers facing young users on Facebook and Instagram, said the agreement largely amounts to "safety theater"—measures designed to appear protective without meaningfully reducing exposure to harmful content or algorithmic manipulation.
Bejar's scepticism carries particular weight given his role in the legal proceedings. During testimony before a judge and jury panel, he detailed how Meta systematically downplayed the prevalence of teen harm on its platforms, deliberately steered young users toward content known to damage self-esteem, and inadequately responded to reports of child exploitation. These allegations formed the cornerstone of the states' case, yet the settlement framework addresses them only peripherally, focusing instead on expanded parental oversight tools and usage monitoring.
The settlement, announced by California Attorney General Rob Bonta and other state officials, requires Meta to pay up to $18 billion over the coming decade alongside implementing new safeguards for teenage users. Bonta declared the deal would "make social media less dangerous for our kids," signalling confidence that enforced remedies represent substantive progress. However, this optimism stands at odds with Bejar's assessment that Instagram and Facebook usage will decline marginally while the underlying safety challenges persist unchanged.
Central to Bejar's criticism is the recognition that several settlement provisions rehash measures Meta previously examined and deliberately rejected. A striking example involves hiding like counts and reaction metrics on posts—a safeguard designed to reduce harmful social comparison among teenagers. Meta's internal research, codenamed Project Daisy, tested this intervention starting in 2019 after researchers flagged concerns that algorithmic content feeds were damaging teen mental health. Company testing demonstrated the change would produce only modest improvements in user well-being while reducing daily active users by approximately 0.09 percent—deemed commercially inconsequential. Rather than implementing the change platform-wide, Meta relegated it to a voluntary opt-in feature, effectively rendering it invisible to most younger users.
The settlement now mandates this same feature, raising uncomfortable questions about whether legally enforced implementations will achieve outcomes that voluntary adoption could not. Internal Meta communications from 2020 show developers explicitly acknowledged they "didn't observe movements in overall well-being measures" when like counts were hidden. This documentation suggests the company has long understood that cosmetic adjustments to interface design cannot address the fundamental business model underlying its platforms—engagement maximisation through algorithmic curation and personalised feeds designed to capture and extend user attention.
Meta's handling of content moderation shortcomings similarly illustrates the gap between alleged harms and settlement remedies. Bejar's oversight of internal research revealed that users experienced negative encounters on Meta platforms at substantially higher rates than the company publicly acknowledged. Leaked research from Meta employees has documented the platform's algorithms' tendency to amplify fitness and beauty content to teenagers already struggling with self-image concerns, while deliberately structuring features to extend session durations beyond what users would independently choose. These architectural decisions, rooted in Meta's advertising-dependent revenue model, receive only cursory attention in the settlement framework.
Instead, the agreement emphasises parental control mechanisms, daily time limits on Instagram and Facebook, and enhanced efforts to identify and remove accounts operated by adults misrepresenting their age. While addressing certain forms of abuse, these measures sidestep the algorithmic recommendation systems and feed curation that Bejar and other critics identify as the core problem. The settlement's presumed reliance on age verification also carries proven limitations—Australian regulators implementing a blanket ban on social media for under-16s have found that eight in ten younger teenagers remain active on platforms, suggesting that technical and policy-based solutions cannot overcome teenagers' determined engagement with these services.
Expert reactions to the settlement among clinicians and researchers who specialise in adolescent mental health have been notably divided. Dr. Jane Conron, a clinical psychologist at Northwestern University's Feinberg School of Medicine, expressed particular concern about one provision requiring Meta to offer a non-algorithmically curated feed as an opt-in feature. Drawing on her clinical experience, Conron predicted this option would remain largely unused, as teenagers gravitate toward algorithmically personalised content regardless of documented harms. She noted that some of her younger patients exhibit such acute emotional dependency on social media platforms that they experience genuine distress when parents attempt to enforce usage restrictions, suggesting that built-in daily time caps could provide meaningful intervention points for families.
While Conron remained cautiously optimistic about certain settlement elements—particularly daily usage caps that could interrupt compulsive engagement patterns—she acknowledged the agreement's failure to require Meta to admit fault or accept responsibility for building inherently harmful products. The settlement's significance, from her perspective, lies not in specific technical remedies but in catalysing broader cultural recognition of social media's documented dangers. She believes the legal action and resultant settlement may fundamentally shift conversations between parents and teenagers about platform use, potentially empowering families to establish boundaries even without platform-enforced restrictions.
For Malaysian and Southeast Asian audiences, this settlement offers instructive lessons about platform accountability beyond American borders. Meta's business model and algorithmic systems operate uniformly across markets, meaning teenagers in Malaysia, Singapore, and throughout the region encounter identical recommendation algorithms and engagement-maximising features that American regulators now scrutinise. The settlement's perceived limitations—that it allows substantial aspects of the current system to continue functioning—suggest that similar regulatory approaches in Asia-Pacific nations may encounter comparable implementation challenges. Regulators across the region increasingly face pressure to address youth mental health concerns linked to social media, yet the American settlement demonstrates that negotiated remedies, however substantial their financial component, struggle to address the fundamental tension between platforms' advertising-dependent business models and genuine teen safety protections.
The broader implication extends to questions about regulatory effectiveness when enforcement relies on negotiated settlements rather than structural reform. Meta's willingness to pay $18 billion across a decade reflects a calculation that financial penalties, even at substantial scale, represent acceptable costs for maintaining operational models that prioritise user engagement and advertising revenue over adolescent well-being. This calculus affects not only American teenagers but young people throughout the world who use Meta's platforms. Lawmakers across Asia-Pacific, observing the American settlement's apparent limitations, face difficult choices about whether alternative regulatory frameworks—mandatory algorithmic transparency, algorithmic feed alternatives, explicit age-based restrictions on algorithmic curation—might prove more effective than negotiated compliance arrangements.
Bejar's continued public advocacy despite the settlement suggests that advocates for tech accountability view this moment as merely one phase in an ongoing effort rather than a conclusive resolution. His characterisation of the agreement as codifying safety theatre reflects a sophisticated understanding that platform companies can adapt to regulatory requirements while preserving the architectural features most consequential to engagement metrics and revenue growth. For regulators across the region evaluating how to protect their younger populations from algorithmic harms, the American settlement's cautionary example underscores that regulatory frameworks must target fundamental design practices rather than peripheral features if they hope to achieve meaningful improvements in teen mental health outcomes.
