Take It Down Act

what is the take it down act? the us federal take it down act requires online platforms to remove non consensual intimate imagery (ncii) — including ai generated or deepfake intimate imagery — within 48 hours of a valid report nima's flexible, rule based approach lets you configure compliance with regulations like this one you can meet the 48 hour deadline by combining two existing building blocks — a queue with a strict sla (see queues docid\ azsiick6eehflzwytql v ), and a queue routing rule that isolates ncii reports into it ahead of everything else (see rule configuration docid\ hfx29ae9o5ktdj2 dokxb ) recommended setup create a dedicated queue for take it down act reports, with its sla set to 48 hours (or tighter, to leave a buffer for review) see queues docid\ azsiick6eehflzwytql v for how to create and configure a queue create a queue routing rule that matches on the input labels covering ncii, and assign it to that queue prioritise the rule above your other queue routing rules , so an ncii report is caught by this rule first, even if it would also match a broader rule (e g a general cyber violence rule) see "how to prioritise rules in nima" in rule configuration docid\ hfx29ae9o5ktdj2 dokxb this aggregates every report carrying the 48 hour legal deadline into one queue with visible, trackable sla countdowns, instead of relying on manual triage recommended input labels to match on reports submitted through the illegal content report form docid\ ic7tcb ig zpyofliekr carry two input labels the category and sub category of illegal content the reporter selected we recommend matching your routing rule on the sub categories that typically constitute ncii cyber violence → non consensual (intimate) material sharing, including (image based) sexual abuse , and non consensual sharing of material containing deepfake or similar technology cyber violence against women → the same two sub categories csam / protection of minors → child sexual abuse material , and csam containing deepfake or similar technology these aren't the only combination that can satisfy the requirement — adjust the list to match your own risk assessment and legal advice add whichever sub categories you land on as input label criteria (or'd together) on the routing rule the rule's source condition must be set first by default, the input label criteria only offers a handful of sub categories (e g phishing , defamation ) the rest — including every ncii sub category above and the cyber violence against women category itself — only appear once you set source to user report illegal content add that condition before selecting input labels, or you won't be able to build a rule that reliably catches these reports