Meta’s agreement is being cast as both a breakthrough for children’s safety and an opening salvo: advocates want the settlement to force industry-wide change, while critics of sweeping regulation warn that social media and AI are far less straightforward targets than tobacco.

The pressure had already been building. In New South Wales, Education Minister Prue Car ordered an August review of AI in schools and asked regulators to consider pausing unsupervised take-home assessments. Her argument was blunt: AI may assist learning, but it must not replace “thinking, creativity and diligence.”¹

This week, Meta agreed to pay up to $17.1 billion and make major changes to Facebook and Instagram after claims that its platforms endangered children through addictive design.² The settlement requires daily usage limits, nighttime blocks for teen users, stronger age-assurance measures and additional parental tools, with an independent auditor monitoring compliance for five years.³ Age assurance is central because even robust safeguards fail if a platform cannot reliably identify its younger users.⁴

For Mike Moore, the former Mississippi attorney general who helped secure the 1998 tobacco settlement, Meta’s deal is a template rather than a final destination. “The settlement with the states is a great first step,” he said, backing a national education and prevention fund instead of a patchwork of local remedies.³ California Attorney General Rob Bonta struck the same note, calling the agreement “a floor conceptually, not a ceiling.”³

But the tobacco analogy has limits. Carnegie Mellon’s Jonathan Caulkins noted that “cigarettes are cigarettes,” whereas the harms and benefits of social platforms and AI may look radically different in five, 10 or 15 years.³ Technology investor Allison Ball similarly warned policymakers to stay “maniacally focused on what could actually help,” rather than embrace rules with unintended consequences.³

The settlement’s shadow is already stretching toward chatbots. Stanford privacy fellow Jennifer King called it “definitely a warning for the AI companies,” as state-level cases and regulation move faster than Washington.⁵ That urgency collides with an AI market still sprinting ahead: Z.ai revealed that the mystery model Ox Alpha was its GLM-5.3-Flash, a release promoted as imminent by AI observers on X.clem 🤗 (ClementDelangue)⁶ The regulatory challenge is no longer just policing yesterday’s social feed; it is keeping pace with tomorrow’s products.