ResearchThursday 14 May 2026via London School of Economics and Political Science

LSE research documents platform safety regression as major platforms replace default protections with optional parental controls

Guardii Analysis

A study published in May 2026 by the LSE Digital Futures for Children centre and 5Rights Foundation documented 108 child safety and privacy design changes across 70 platforms between 2024 and 2026, finding that the four largest platforms, Meta, Google, TikTok and Snapchat, have shifted away from protective by-default design changes toward optional end-user tools such as parental controls, reversing the trend identified in the first phase of the research. The report, tracking platform responses to the UK Age Appropriate Design Code, the UK Online Safety Act and the EU Digital Services Act, found that significant child safety risks including compulsive engagement and ongoing content harms continue across many platforms despite existing regulatory measures, and that only five of the platforms contacted directly responded to information requests about their child safety implementations.

The documented shift from safety by default to opt-in parental tools exposes the structural weakness that targeted, intelligent detection systems are designed to close. Guardii, backed by Startmate and operating as a Meta Business Partner, does not rely on a parent remembering to activate a control, a child choosing not to disable a setting, or a platform volunteering transparency about the efficacy of its measures. Instead, its real-time detection architecture operates continuously across the platforms children actually use, identifying escalation toward grooming, sextortion, CSAM sharing, cyberbullying, and acute distress in the conversation itself, surfacing a child at genuine risk to the adult or professional responsible for their welfare without handing that adult a feed of everything the child has said online. Where regulation sets a floor and voluntary platform measures depend on user action, pattern-based AI detection intercepts harm at the moment a predatory adult makes contact, translating the policy goal of child protection into an operational system that functions whether or not a parent has navigated a settings menu or a teenager has chosen to disclose that they are in danger.

Related Articles