How can platforms self-govern as social media bans spread
Last updated: 4 October 2026
A board can count how many underage accounts its platform has removed yet still struggle to explain whether the experience has become safer.
Platforms can govern themselves more credibly by defining the outcomes they want, measuring the relationships their services encourage, and testing whether safeguards actually change those outcomes.
What do the new restrictions actually require?
Australia’s restrictions have been in force since 10 December 2025, requiring covered platforms to take reasonable steps to prevent under-16s from creating or keeping accounts. Australia’s eSafety Commissioner also records enforcement amendments passed in September 2026, strengthening its ability to investigate compliance.
In the US, Florida’s legislation prohibits under-14s from becoming account holders on platforms meeting its statutory criteria and requires parental consent for 14- and 15-year-olds. The law remains subject to litigation, so enactment and a final judicial determination should not be confused.
Europe is pursuing several approaches. On 7 September 2026, Denmark announced a proposal for a 15-year social media age limit, with implementation expected in July 2027. Alongside national initiatives, the European Commission’s guidelines under the Digital Services Act recommend changes to recommender systems and defaults for features associated with excessive use. Furthermore, on 15 June 2026, the UK government announced a ban on social media use by children under the age of 16. The announcement followed a consultation earlier this year on improving children’s relationship with mobile phones and social media. The ban is expected to be in place from spring 2027.
These measures differ in scope and legal status. Together, they put a practical question before platform leaders: what evidence would demonstrate that the service is being operated responsibly?
What should platforms measure beyond engagement?
Consider two teenagers spending an evening in a gaming community. One is organising a tournament with friends; the other keeps returning to content they find upsetting. The same amount of time online could describe very different experiences. This hypothetical example illustrates why duration needs context.
Fandom makes that distinction particularly important. Returning to a creator, discussing a story or contributing fan art can express enjoyment and belonging. A governance system should preserve those opportunities while examining patterns that warrant attention, without treating enthusiasm itself as evidence of harm.
A useful measurement framework would combine behavioural patterns over time with user feedback and contextual evidence. The aim would be to distinguish voluntary participation, trust and community connection from possible loss of control or unwanted influence. No individual click, return visit or score should settle that judgement.
For boards, the practical questions could look like this:
Governance question: Evidence to examine
Can users control their experience?
Use of feed controls, blocks and mutes, plus reported control.
Are safeguards changing outcomes?
Changes in exposure and user feedback after an intervention.
Are benefits and risks unevenly distributed?
Differences between age groups and communities, with uncertainty reported.
These are proposed assessment questions, rather than a universal safety standard. Their value depends on clear definitions, appropriate data and testing against what users actually experience.
How can measurement become accountable action?
Platforms should begin with a specific concern, such as repeated unwanted recommendations, then define a safeguard and an outcome they can assess. They might change a recommendation rule, notification default or user control and examine the results over a defined period.
Where ethical and appropriate, comparisons can help distinguish an intervention’s effects from changes elsewhere on the service. Platforms should avoid experiments that knowingly increase children’s exposure to harm. Reviews should also consider unintended consequences, including whether protections make supportive communities harder to access.
The board’s role is to require ownership, escalation thresholds and a record of decisions. Consumer groups and young people should help interpret the results, while independent scrutiny can challenge assumptions. Any behavioural inference should use proportionate data collection, privacy protections and a route to question consequential decisions.
Where could Thetafan Metrics contribute?
Thetafan is developing relationship intelligence built on its proprietary, patent-pending technology. Its starting point is that interactions become more informative when considered together over time, helping organisations understand relationships around content, creators and communities.
The proposed governance application is to support measuring changing relationships, identifying patterns that require attention, and assessing interventions. This development work sits alongside Thetafan’s Metrics offering; its effectiveness for child-safety decisions would need application-specific validation. It should not be presented as a clinical diagnosis or a guarantee of regulatory compliance.
Measurable self-governance means making platform decisions open to examination. What changed, why it changed and what evidence followed. Relationship metrics could contribute to that evidence, provided their limits are visible and people remain accountable for the decisions.