Meta Platforms has agreed to pay up to $18 billion over the next decade to settle child safety allegations brought by US state attorneys general, but a former company safety engineer at the centre of the case has cast serious doubt on whether the accord delivers meaningful protections for young users. Arturo Bejar, who testified about Meta's internal knowledge of harms inflicted on teenagers through its platforms, said the settlement essentially formalizes what he calls "safety theatre"—cosmetic changes that leave the underlying business model and algorithmic systems intact.
Bejar's concerns carry weight because his testimony formed a crucial part of the states' legal case against Meta. During proceedings last week, he detailed how the company systematically underreported the prevalence of harm affecting teens, deliberately exposed them to self-esteem-damaging content, and failed adequately to investigate child sexual exploitation reports. These allegations struck at the heart of Meta's responsibility as a platform host, suggesting not merely negligence but deliberate choices that prioritized engagement metrics over youth wellbeing.
The settlement announced by California Attorney General Rob Bonta and fellow state officials presented a starkly different narrative. Bonta declared the agreement would "make social media less dangerous for our kids," framing it as a watershed moment in protecting children online. Meta, for its part, characterised the deal as consistent with its "longstanding efforts to empower parents and support teens," emphasising what it described as real and enforceable safeguards. Yet Bejar's assessment—delivered directly to the judge and jurors who heard the case—suggests the accord may represent less a victory for child safety than a carefully negotiated compromise that allows Meta to continue largely as before.
Among the settlement's remedies is a commitment by Meta to hide like and reaction counts on posts, a measure designed to reduce what psychologists term "negative social comparison" among teenagers. Revealing the gap between settlement promises and company priorities, internal documents show Meta considered this exact change years earlier. In 2019, researchers explored what they called "Project Daisy," investigating whether hiding likes would improve teen wellbeing. The results underwhelmed: a 2020 slide presentation to Mark Zuckerberg noted that well-being measures showed no meaningful improvement, and business modelling suggested daily user numbers would decline by only 0.09% if implemented.
This history exposes a troubling pattern. Remedies secured through legal settlement and state pressure represent measures Meta's own researchers had already tested and internally rejected as insufficiently consequential. The company ultimately offered like-hiding as an opt-in feature—essentially a permission to ignore it rather than a universal change. For Bejar and others tracking Meta's accountability, this precedent suggests the settlement perpetuates a cycle where the company adopts low-impact reforms while maintaining the algorithmic architecture and engagement-maximising design that researchers themselves flagged as problematic.
Meta's internal research, leaked by employees or extracted through legal discovery, has documented specific harms that the settlement addresses only peripherally. Company algorithms were shown to concentrate high volumes of fitness and beauty content toward teenagers already struggling with self-esteem, a form of targeted exposure to content likely to worsen body image concerns. Design features were intentionally engineered to extend browsing duration beyond what users would independently choose, exploiting psychological vulnerabilities common in adolescence. Bejar's own oversight of safety metrics revealed that teenagers experienced problems on Meta's platforms at rates substantially exceeding the company's public claims—a discrepancy suggesting either analytical shortcomings or deliberate understatement.
The settlement's actual provisions pivot toward parental control mechanisms, time-use restrictions, and enhanced detection of minors falsely claiming adult status. While these tools may offer marginal improvements for conscientious parents, they sidestep the core concern animating Bejar's testimony: that Meta's fundamental business incentives drive it to prioritise addictive engagement over user wellbeing, and that algorithmic recommendation systems encode these priorities into every interaction. A time limit or parental dashboard does not alter the algorithmically-curated feed itself or Meta's financial dependency on maximising user attention.
The Australian experience offers cautionary perspective. After banning social media use by under-16s, regulators discovered that approximately eighty percent of young teens remained active on platforms, circumventing the law through age misrepresentation. Meta's proposed settlement relies substantially on age verification and minor-detection improvements, yet the Australia case demonstrates the weakness of such technical approaches when enforcement depends on user self-reporting. The company's systems remain vulnerable to determined adolescents, and even if age verification improved, this addresses only whether minors access platforms, not whether the platforms themselves have become safer.
Mental health specialists offered mixed assessments of the settlement's likely impact. Dr Jane Conron, a clinical psychologist at Northwestern University's Feinberg School of Medicine, highlighted that the settlement's requirement for non-algorithmic feed options faces a significant practical barrier: it operates on an opt-in basis, suggesting most young users will never discover or enable it. However, she identified the daily time-cap on Instagram and Facebook usage as potentially meaningful, particularly for teenagers whose emotional dependence on these platforms creates crisis when access is restricted. Some of her young patients experience acute distress when parents attempt to limit screen time, suggesting that built-in limits might alleviate both the psychological struggle and the family conflict surrounding such restrictions.
Conron also acknowledged that the settlement's very existence, regardless of its technical limitations, may catalyse broader cultural reassessment of social media's role in adolescent life. Although Meta avoided admitting fault or conceding that its products inflict harm, the settlement signals public and legal recognition of genuine dangers. This framing may shift conversations between teenagers and parents away from simple usage rules toward deeper discussion of how platforms operate and what they do. From this perspective, the accord's value lies not in its specific technical remedies but in legitimising concerns that Meta previously downplayed and establishing a legal precedent that holds platforms accountable for youth safety impacts.
For Malaysia and the Southeast Asian region, the Meta settlement carries implications extending beyond the specific terms and broader debates about US regulatory power. The tech ecosystem across Southeast Asia increasingly mirrors US patterns, with platforms like TikTok, Instagram, and Facebook dominating teen engagement while facing minimal regulatory scrutiny. Local regulators lack both the resources and legal frameworks available to US state attorneys general. The Meta case, despite Bejar's reservations, establishes that accumulated evidence of platform harm—particularly when coupled with internal company documents contradicting public safety claims—can generate consequences. Yet the settlement also illustrates how sophisticated companies can navigate accountability through negotiated settlements that impose costs without requiring fundamental business model changes. Regional policymakers observing these dynamics face a choice: invest in robust regulatory capacity and legislative frameworks now, or accept that future accountability efforts will likely produce comparably modest outcomes.
