Frontier Model Forum
- Type
- Standards body
- Lieu
- International — International (industry consortium, US-based)
- Dernière vérification
- 2026-09-22
- Prochaine vérification
- 2027-03-21
Pas encore traduit — affiché en anglais.
Comment les joindre
- General enquiries
- Contact page
- Membership criteria (for frontier developers only)
Ce canal est conservé uniquement comme référence. Sa vérification est absente, date de plus de 180 jours, ou la période de contact n'est pas ouverte actuellement ; ne vous y fiez pas avant une nouvelle vérification.
- Homepage
Ce qu'il fait
Industry-supported non-profit founded by Amazon, Anthropic, Google, Meta, Microsoft and OpenAI. Publishes technical reports and issue briefs on frontier capability assessments and safety frameworks, with workstreams on AI-Bio, AI-Cyber, AI-Nuclear, AI Security and Safety Frameworks. Runs information-sharing between member labs and with government.
Évaluation franche
Effectively closed. It is a members' club of the largest labs with a single generic inbox. Sending a concern here is close to sending it to the labs' shared comms function. Its technical publications are worth reading; it is not a route in.
Préoccupations couvertes
Severe misuse risk at the frontier — CBRN and offensive cyber uplift — plus security of model weights. Explicitly narrow: it is a misuse-and-security body, not an alignment or near-term-harms body.
Canal gouvernemental
Strong and quiet — the FMF is a recognised industry interlocutor for the US and UK governments and for the AI Safety/Security Institute network.