Français

Organismes ayant un pouvoir

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 · Email address
    Vérifié · 2026-09-22

    The only published address

  • Contact page · Web form
    Vérifié · 2026-09-22

    General

  • Membership criteria (for frontier developers only) · Official membership list
    Vérification expirée · 2026-09-22

    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.

    Organisations that develop and deploy frontier models and have a safety framework

    Individuals cannot join.

  • Homepage · Homepage
    Non contrôlé

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.

Sources

Une erreur ?