Arturo Bejar, an engineer who previously worked on safety at Meta Platforms, has cast doubt on the effectiveness of a major settlement announced between the social media giant and a coalition of American state authorities. Bejar's courtroom testimony, delivered last week before a judge and jury, detailed systematic failures in how Meta protected young users, including the deliberate underreporting of harm incidents, algorithmic promotion of self-damaging content, and inadequate response to child predation complaints. Now, following Wednesday's announcement of an settlement framework, Bejar contends that the agreement does little to remedy the core issues he exposed, instead representing what he characterises as a continuation of superficial safety measures.
The settlement, which will see Meta pay up to $18 billion over the coming decade to participating states, includes provisions designed to restrict teenage usage patterns on Facebook and Instagram, alongside enhanced parental oversight tools. California Attorney General Rob Bonta framed the deal as a meaningful step toward making social media demonstrably safer for children, emphasising that the agreement introduces enforceable changes, improved transparency, and concrete protections. Meta itself responded by stating that the settlement builds upon existing efforts to grant parents greater authority and assist teenagers in using its platforms responsibly.
Bejar's scepticism stems from a detailed understanding of Meta's internal operations and decision-making processes. He argues that many remedies included in the settlement represent measures the company had previously examined and deliberately rejected, having concluded they would produce minimal benefits either for user wellbeing or for Meta's commercial interests. This assessment is backed by leaked internal documentation that emerged during legal proceedings, revealing how Meta employees and researchers weighed the costs and benefits of proposed safety changes.
A particularly instructive example involves the hiding of social media engagement metrics. Meta has committed under the settlement to conceal likes and reaction counts on user posts as a mechanism to reduce unhealthy social comparison among teenagers. However, internal research dating to 2019, when Meta considered implementing this same feature, suggested the impact would be modest at best. Project Daisy, as Meta's internal testing programme was named, found that obscuring like counts produced only marginal improvements in user wellbeing measures. A 2020 presentation to Mark Zuckerberg noted that researchers had observed no meaningful movement in overall wellbeing indicators following the test implementation. The business implications were equally insignificant: simulations predicted daily active users would decline by approximately 0.09 percent if the change became permanent. Rather than implementing a mandatory shift, Meta instead offered users the option to hide likes voluntarily—an approach that limited adoption and impact.
The settlement's approach reflects a broader pattern in how Meta's safety obligations are being addressed. Rather than tackling the algorithmic recommendation systems that Meta's own researchers have documented as problematic, the agreement emphasises parental controls, screen time limits, and better detection of age misrepresentation. Internal research flagged by Meta's teams has shown that platform algorithms disproportionately promote fitness and beauty content to teenagers already struggling with self-esteem, and that feature design deliberately extends user engagement beyond what individuals would choose independently. These fundamental architectural issues remain largely unaddressed by the settlement framework.
Bejar's original testimony centred on Meta's underestimation of harm prevalence on its platforms. His work at the company involved research demonstrating that teenagers experienced negative outcomes at rates substantially higher than Meta's publicly reported statistics indicated. The discrepancy between internal prevalence estimates and official communications formed a cornerstone of the states' legal case. Yet the settlement accords only cursory treatment to these findings, instead focusing on the peripheral mechanisms of parental visibility and usage time.
The question of age verification presents a particularly stubborn implementation challenge. The settlement includes measures intended to better identify and remove underage users from platforms, but enforcement depends largely on self-reported age information provided at account creation. This dependence on user honesty has proven ineffective in practice. Australia's experience offers an instructive precedent: following the government's ban on social media access for users under 16, authorities discovered that approximately eight in ten young teenagers remained active on these platforms, prompting consideration of more stringent enforcement mechanisms. Meta faces similar verification challenges that the current settlement does not substantially resolve.
Professional responses to the settlement framework have been mixed, particularly among experts focused on adolescent mental health. Dr. Jane Conron, a clinical psychologist at Northwestern University's Feinberg School of Medicine, noted that one settlement provision requires Meta to offer algorithmically unfiltered feeds to teenagers on an opt-in basis. Conron predicted this feature would attract minimal voluntary adoption, as most users lack incentive to abandon the engagement-optimised algorithmic experience. However, she identified one potentially meaningful component: daily usage caps on Instagram and Facebook that would function automatically rather than relying on user discipline.
Conron's clinical experience demonstrates the psychological intensity surrounding these platforms for some adolescents. She reported instances where young patients exhibit acute emotional distress when parents attempt to restrict application usage, sometimes reaching the point of tears when facing reduced access. A built-in usage ceiling, Conron suggested, might provide necessary friction without placing the enforcement burden entirely on parents and teenagers negotiating household restrictions. Beyond individual benefit, she posited that the settlement's very existence might shift cultural conversation, forcing greater acknowledgment among families and teenagers of social media's substantive effects on wellbeing.
The settlement notably stops short of requiring Meta to admit culpability or acknowledge that its products were deliberately engineered to cause harm. This absence of formal liability acknowledgment reflects the negotiated nature of settlement frameworks, where companies typically avoid admissions that might expose them to further legal liability or reputational damage. Conron, however, viewed the settlement's announcement itself as symbolically significant, suggesting that public recognition of the problem's severity might catalyse behavioural shifts independent of the agreement's specific technical provisions.
For Malaysian and Southeast Asian observers, this settlement offers important lessons regarding technology regulation and corporate accountability. Meta's platforms, including WhatsApp and Instagram, command substantial user bases across the region, with Malaysia representing a particularly significant market. The mechanisms through which Meta evaluates safety trade-offs—weighing user wellbeing against business metrics and user retention—operate globally, meaning research and deliberations conducted in San Francisco directly affect teenagers across Asia-Pacific. The gap between documented internal understanding of harms and external public commitments revealed by this case has implications for how regional governments might structure their own oversight frameworks and enforcement approaches.
Moreover, the settlement highlights the structural limitations of monetary penalties in incentivising corporate behaviour change. At $18 billion over a decade, the payment represents a significant sum, yet several analysts have noted that for a company of Meta's scale and profitability, the cost represents a manageable business expense rather than an existential constraint. Bejar's core criticism—that the settlement institutionalises surface-level compliance rather than fundamental product redesign—suggests that financial penalties alone may be insufficient to motivate genuine transformation in how social media platforms prioritise user safety against engagement maximisation.
The fundamental tension exposed by this case concerns the business model underpinning social media platforms globally. Engagement-driven advertising revenue creates inherent incentives to maximise time spent and emotional investment in applications, regardless of documented mental health consequences. Until regulatory frameworks or competitive dynamics force a structural realignment of these incentives, settlements and compliance measures may primarily generate the appearance of safety improvements rather than substantive protections. For teenagers across Southeast Asia and globally, this distinction between safety performance and safety perception may prove consequential.
