Meta just admitted they could've saved your kids all along (overturned.substack.com)

Meta quietly acknowledged it can now identify and proactively block “risky stunts” from teen feeds — a capability the company says was delivered after “several months” of engineering work. That admission comes after years of high-profile tragedies and repeated corporate claims that large-scale moderation was impossible. A former Meta engineer and advocates argue the technology to prevent viral, harmful challenges always existed; legal pressure, lawsuits and congressional scrutiny finally made the company prioritize it. Importantly, Meta says this will be automatic (not reliant on user reports) and targeted at teen experiences, but past teen-safety features have repeatedly underperformed, raising skepticism about rollout, scope and enforcement. Technically and ethically, the announcement matters because it undermines the long-held excuse that recommendation systems can’t be constrained: recommendation and filtering models can be tuned to suppress specific classes of content — including material that exploits adolescent neurobiology (immature prefrontal risk-assessment vs hyperactive social reward circuits) — which platforms’ engagement-optimizing algorithms have amplified. The implications are broad: proactive filtering could be extended to grooming, sextortion, illicit drug distribution and more, but only if regulators require transparency, audits, continuous testing and legal duties of care (e.g., KOSA). Advocates urge verification, independent oversight and binding accountability rather than PR-driven “solutions” retrofitted only after harm and pressure.