Overview
- Meta rolled out a 24/7 AI support assistant on Facebook and Instagram for iOS, Android and desktop Help Centers, offering fast responses to account issues such as reporting scams, resetting passwords and guiding appeals.
- The company plans a staged, multi‑year deployment of advanced AI to police severe violations including terrorism, child exploitation, drugs, fraud and scams, with humans retained for high‑risk calls like law‑enforcement reports and account‑disablement appeals.
- Early tests cited by Meta show the systems mitigating about 5,000 scam attempts per day, reducing reports for the most‑impersonated celebrities by over 80%, and detecting twice as much violating adult sexual‑solicitation content with more than 60% fewer mistakes.
- Meta says the upgraded enforcement can operate in languages used by 98% of people online and can rapidly adapt to cultural nuance, code words, emoji use and slang.
- As these systems scale, Meta will reduce reliance on third‑party moderation vendors; a Reuters report of potential workforce cuts of roughly 20% remains disputed by the company, and law‑enforcement officials have criticized the quality of some automated tips.