Trust & safety — one screen, one write pathi

ENGINE — moderation (MAV-owned map, 2026-08-09): FG’s T&S Overview converges with Folo’s /admin/trust-safety — one surface, both mount it. Every tile is a read that links the queue where the record lives; per-case actions (freeze · release · ban) work in place (§17e) — the write routes via the subject platform’s admin API (one write path, one audit trail). Fraud-controls tabs adopt Folo’s shape (the fuller implementation); the consolidation tiles adopt FG’s. folo.link answer: this page COMPOSES per-platform queues plus the one cross-platform identity read; actions act here and route to the owning platform — the deep-link into the owning queue is the quiet secondary, never the only path.

Fraud controlsi

The risk engine scores four signal types: velocity (spend/msg bursts), device co-occurrence (one device, many accounts), chargeback-proneness, synthetic identity. Freeze ≠ ban: freeze halts earn + spend pending human review and is reversible; ban is terminal. Owner + write path: both live on the account record of the subject’s platform — the platform tag on each case names whose ladder a decision writes; freeze/ban route via that platform’s admin API and are audited there. §17e: they work in place from this consolidation on every skin — the single write path constrains the write, not the operator; the owning queue stays one quiet deep-link away, never the only path. Effect: freeze halts earn + spend on that platform pending review (reversible, fans see content but cannot pay); ban is terminal on that platform — it goes family-wide only if the identity engine ties the same person to other accounts, and that is a separate, audited decision per platform.