Meta Platforms has reached an $18 billion settlement with a coalition of American states over child safety concerns, yet the arrangement has immediately drawn fire from the engineer whose testimony was instrumental in building the case against the social media giant. Former Meta safety officer Arturo Bejar contended that the deal does not substantially tackle the problems he revealed during court proceedings, suggesting instead that it may entrench what he terms "safety theater"—a facade of protection without genuine safeguards.
Bejar's evidence during trial painted a damaging picture of Meta's internal operations. He testified that the company had systematically downplayed how frequently teenagers encountered harmful content on Facebook and Instagram, deliberately steered young users toward material designed to diminish their self-worth, and inadequately investigated reports of child exploitation. These allegations formed the backbone of the states' case, yet the final settlement appears to sidestep many of these central concerns, focusing instead on peripheral remedies that internal Meta research had previously deemed ineffectual.
California Attorney General Rob Bonta hailed the settlement as a landmark that would "make social media less dangerous for our kids," while Meta itself maintained it was "building on longstanding efforts to empower parents and support teens." However, Bejar's subsequent statement suggested a fundamental disconnect between the relief secured and the actual dangers he had documented. He warned that while Instagram usage might decline marginally, the underlying architecture making the platforms unsafe would remain intact.
The settlement's centerpiece involves Meta implementing various restrictions on teenage user experience, including limiting daily time spent on Facebook and Instagram and hiding like counts and reaction metrics from younger users' posts. Yet this latter measure reveals the hollowness at the agreement's core. Meta had considered hiding likes as far back as 2019, when internal research code-named "Project Daisy" tested the concept specifically to address concerns that social comparison was damaging teen mental health. The company's own testing revealed minimal impact: researchers found no meaningful shift in overall well-being metrics, while the business consequences were negligible, with projections suggesting only a 0.09 percent decline in daily active users.
The decision to implement hiding likes only as an opt-in feature meant few users actually adopted it, ensuring Meta had already determined this remedy would prove largely ineffectual before the settlement mandated it. That these measures now appear as victories in a major regulatory accord illustrates how the deal may institutionalize approaches Meta had already vetted and found wanting. The settlement, in essence, codifies what Bejar described as theater rather than substance.
Meta's internal research has documented far more concerning dynamics than those addressed in the settlement. The company's own researchers identified how its algorithms preferentially recommend fitness and beauty content to teenagers already struggling with self-esteem, creating feedback loops that intensify insecurity. Engineers also flagged that Meta deliberately designs features to extend user engagement beyond what young people would naturally choose, effectively engineering compulsive use patterns. Additionally, investigations by Bejar's team uncovered that Meta was significantly underreporting the prevalence of negative user experiences on its platforms compared to official company estimates.
The settlement's remedies disproportionately emphasize parental controls, time limits, and improved detection of adults misrepresenting themselves as minors. Yet these provisions sidestep the algorithmic and design choices that Bejar highlighted as core to the harm. Even the age-verification mechanisms receive skeptical attention: in Australia, where legislation prohibits social media use by those under sixteen, regulators discovered that roughly 80 percent of young teens remain active on platforms, despite self-reporting systems and platform policies. This suggests that technical solutions absent deeper platform redesign may prove equally ineffective.
Mental health professionals remain divided on whether the settlement represents meaningful progress. Dr. Jane Conron, a clinical psychologist at Northwestern University's Feinberg School of Medicine, expressed skepticism about provisions requiring Meta to offer a non-algorithmic feed to teenagers on an opt-in basis, predicting such features would see minimal adoption. However, she identified potential value in daily usage caps, noting that some of her young patients experience emotional distress when attempting to reduce time on Meta's apps. She acknowledged that simply requiring these features might shift broader cultural conversations about social media within families and possibly signal that regulators and the public increasingly recognize the genuine scope of harm involved.
The settlement notably does not require Meta to admit wrongdoing or accept that it built harmful products. Rather, it functions as a financial and operational settlement without acknowledging liability—a structure that allows Meta to frame compliance as voluntary refinement rather than remediation of systemic failures. This distinction matters, as it potentially limits future regulatory leverage and litigation by establishing precedent that massive settlements can be reached without companies conceding fundamental wrongdoing.
For Southeast Asian readers, particularly Malaysian observers, this settlement carries particular significance. Meta's platforms, especially Instagram and Facebook, dominate social media usage across the region, meaning that protections established in the United States often eventually influence standards elsewhere. However, the apparent inadequacy of American regulatory outcomes suggests that regional governments seeking to protect young users may need to pursue independent investigation and potentially more stringent requirements than those contained in this accord. The Malaysian market, with its large youth demographic and high social media penetration, warrants particular vigilance regarding Meta's algorithmic practices and their documented mental health consequences.
The settlement ultimately illustrates the tension between regulatory action and corporate capture of reform processes. While the $18 billion figure appears substantial, spreading over a decade it represents a manageable business expense for Meta—essentially a cost of operations rather than a genuine recalibration of corporate incentives. Bejar's critique suggests that without requiring fundamental changes to algorithmic recommendation systems and engagement design mechanisms, regulatory settlements may produce paperwork compliance rather than protection. The question now facing policymakers globally is whether this settlement represents a template for future agreements or a cautionary example of how financial penalties and marginal policy adjustments can obscure the persistence of structural harms.
