Ofcom chief Melanie Dawes warns social media companies to prove their algorithms protect under-18s from seeing harmful content, or face enforcement action
Ofcom chief Melanie Dawes reveals she held meetings with US AI firms over Online Safety Act — Tech companies will be subject …
Context & Ripple Effects
Ofcom’s warning advances a regulatory arc from its proposed algorithm changes to reduce children’s exposure to toxic material to the July 2025 child-safety measures that put platforms on notice of substantial penalties. The focus is shifting from publishing safety duties to demonstrating that recommendation systems meet them.
Dawes’s meetings with US AI firms broaden the issue beyond conventional social platforms: algorithmic safety is becoming an implementation and assurance question for services that may shape what young users encounter.
First-order effects
- Social-media companies must be prepared to evidence how their algorithms limit under-18s’ exposure to harmful content, rather than rely on policy statements or high-level safety commitments.
- Ofcom gains a clearer basis to test platforms’ child-protection controls and escalate to enforcement where their evidence or safeguards fall short.
Second-order effects
- Product, trust-and-safety, and compliance teams will need to connect age-assurance signals with recommendation and moderation controls, reinforcing the operational demands set by Ofcom’s child-safety measures.
- US AI firms engaging with Ofcom face pressure to account for how their systems may be used by younger people in the UK, extending the regulator’s practical enforcement surface beyond established social networks.
Third-order effects
- If enforcement follows, UK online-safety compliance will increasingly turn on auditable system behavior—how ranking, recommendation, and access controls work for minors—not simply removal of individual pieces of content.
- The likely durable shift is toward age-aware service design, although its effectiveness will depend on whether age checks and algorithmic safeguards can resist circumvention while avoiding excessive collection of user data.
The trend: Child online-safety regulation is moving from content rules toward demonstrable, age-aware controls over the algorithms and systems that distribute content.