The policy and systems layer of platform safety — the full taxonomy of what gets enforced, how severity is determined, and when a case goes beyond standard review into specialist escalation.
The full landscape of policy violations I enforced — organised by category. Each type required a different assessment lens, threshold, and enforcement response.
Self-harm content required a three-tier assessment — the same piece of content could warrant very different responses depending on intent, framing, and immediacy of risk.
Not every violation warrants the same response. Enforcement was graduated — matched to the severity of the violation, account history, and platform context.
Most cases were handled directly. A small subset required escalation — where severity, legal risk, or specialist knowledge exceeded standard review. Knowing when to escalate was as important as the decision itself.