AI Policy Wiki
Dashboard

Frontier Model Forum

medium confidence · updated 2026-06-08

Industry body for frontier AI safety research and policy, founded 2023 by Anthropic, Google DeepMind, Microsoft, and OpenAI.

The Frontier Model Forum is an industry body for frontier AI safety research and policy, established in July 2023 by Anthropic, Google DeepMind, Microsoft, and OpenAI. It operates as a venue for industry-led AI safety coordination, working through groups on technical safety research, the sharing of safety evaluations, and engagement with governments and policymakers.

FieldValue
TypeIndustry body / nonprofit
FoundedJuly 2023
Members (founding)Anthropic, Google DeepMind, Microsoft, OpenAI
Members (added)Amazon, Meta (May 2024)
Executive DirectorChris Meserole
MissionAdvance AI safety research; develop standards; facilitate information sharing among frontier labs

Overview

The Frontier Model Forum was established in July 2023 by the four largest frontier AI developers — Anthropic, Google DeepMind, Microsoft, and OpenAI — as a venue for industry-led AI safety coordination. Its stated mission is to advance AI safety research, develop standards, and facilitate information sharing among frontier labs. It operates through working groups on technical safety research, sharing of safety evaluations, and engagement with governments and policymakers. Chris Meserole serves as its executive director (Source: frontiermodelforum.org).

The Forum emerged in the same period as the Bletchley Declaration (Nov 2023) and the Seoul Frontier AI Safety Commitments (May 2024), and represents the industry-side institutional infrastructure for the voluntary safety coordination that these summits requested.

Membership and activities

Amazon and Meta joined the Forum on May 20, 2024, expanding membership to six firms alongside founding members Anthropic, Google, Microsoft, and OpenAI; the Forum requires members to show that they are actively building toward or deploying frontier models and have a track record of building safety mitigations into development (Source: frontiermodelforum.org). The Forum operates an AI Safety Fund that issues research grants — it announced a new set of grantees on December 11, 2025 — and runs an information-sharing program among members on frontier AI threats and vulnerabilities, on which it published a progress update on February 16, 2026 (Source: frontiermodelforum.org; frontiermodelforum.org).

Role in AI governance

The Frontier Model Forum is a venue through which frontier AI labs collectively engage on safety standards. It sits between individual company responsible scaling policies, such as Anthropic's RSP, and formal legislation, and has been described as a candidate vehicle for coordinated safety commitments.

The Forum functions as the industry counterpart to the international AI safety summit processes. It is the body governments have pointed to as a channel for coordinating safety commitments across the G7 Hiroshima Code of Conduct for Advanced AI (2023), The Bletchley Declaration (AI Safety Summit, 1–2 November 2023), and Frontier AI Safety Commitments (Seoul, 2024) processes. IMDA Singapore references the Forum in its governance context alongside the Hiroshima Code, the EU AI Act, and the NIST AI RMF, citing it as a channel for industry input to multilateral AI governance.

Limitations and criticism

The Frontier Model Forum is a voluntary body with no enforcement mechanisms. Critics argue it primarily serves to coordinate industry messaging and defer binding regulation. Its membership is limited to six large US frontier developers and does not include major Chinese labs, Mistral, or smaller safety-focused developers.

Relationships