Facebook tightens its policies around self-harm, suicide, and eating disorder content to prevent its spread, after consulting with experts
Context & Ripple Effects
This policy tightening caps a four-year arc that started as a support story: Facebook first linked troubled posters to trained experts in 2015, added faster moderator flagging and self-help guides in 2016, then extended detection to Live video and Messenger with AI spotting at-risk users before any report arrives.
The shift from helping users to restricting content follows sustained criticism of how those systems behave in practice — experts argued earlier this year that Facebook's opaque practice of flagging possible suicide threats directly to police may be inaccurate, ineffective, or unsafe, and its AI detection rollout skipped the EU entirely. Consulting outside experts before rewriting the rules reads as a response to that legitimacy gap.
First-order effects
- Users posting or searching for self-harm, suicide, or eating disorder material hit tighter enforcement immediately, and Facebook's moderation teams gain explicit rules for content categories — notably eating disorders — that the earlier tooling never explicitly governed.
Second-order effects
- Suppression cuts into the very signals Facebook's own detection stack depends on: the AI built since 2017 to spot suicidal posts trains on user-generated distress content, so stricter removal narrows the dataset even as it shrinks exposure.
- Rival platforms face pressure to match the tightened bar, since a visible gap in self-harm or pro-eating-disorder handling becomes an easy criticism point for the same experts Facebook just consulted.
Third-order effects
- If the pattern holds, platform safety policy migrates from reactive support prompts toward proactive content restriction validated by outside experts — a governance model regulators could cite when demanding equivalent diligence, particularly in markets like the EU where Facebook already withheld its AI tools.
The trend: Social platforms are moving suicide and self-harm work upstream — from detecting distressed users to suppressing harmful content at the source, with expert consultation replacing internal judgment as the source of legitimacy.