Field builder · Policy · NY on public filings) · Est. 2023
Safe AI Forum (SAIF) is a US 501(c)(3) nonprofit that aims to foster international coordination to minimize extreme AI risks and ensure AI benefits are shared by all. SAIF’s institutional “field-building” mechanism is to convene cross-jurisdiction scientific and policy communities, then translate the resulting consensus into concrete governance concepts, research agendas, and stakeholder action.
Its flagship program is the International Dialogues on AI Safety (IDAIS), a series of international dialogues whose outputs include public consensus statements and policy guidance intended to inform domestic and international governance approaches. Operationally, SAIF combines (1) convening (IDAIS dialogue events across major AI jurisdictions), (2) research and policy translation (e.g., the IDAIS Policy Guide mapping dialogue consensus to policy levers), and (3) advisory/workshop support for aligned organizations. SAIF positions its work around shared understanding and collaboration among major AI actors, emphasizing governance coordination as a prerequisite for reducing catastrophic risks from frontier capabilities. SAIF also reports that it does not accept corporate funding for IDAIS (as described in its IDAIS Policy Guide).
No curated programs yet. Check the careers page ↗.
Publications are not yet indexed for this organization. Read them on safeaiforum.org ↗.
SAIF announced that, effective May 2, 2025, it transitioned to independent 501(c)(3) nonprofit status and ended fiscal sponsorship with FAR.AI.
Safe AI Forum spin-out from FAR.AI; transition to independent nonprofit operationsFAR.AI published an announcement describing the Safe AI Forum’s transition from a fiscally sponsored project to an independent non-profit and noting board continuity via Adam Gleave’s appointment (as described by FAR.AI).
IDAIS-Venice consensus statement published (AI safety as a global public good)The IDAIS-Venice dialogue is documented with dates September 5–8, 2024 and includes a public consensus statement on AI safety as a global public good and the need for international governance coordination.
IDAIS-Beijing dialogue output: red lines in AI development and enforcement roadmapThe IDAIS-Beijing dialogue (March 10–11, 2024) documents consensus “red lines” for AI development (including restrictions around autonomous replication/improvement, power seeking, WMD assistance, cyberattacks, and deception) and a roadmap for red-line enforcement.
IDAIS-Oxford inaugural dialogue documented (global action on AI safety research and governance)The IDAIS-Oxford page documents the inaugural IDAIS dialogue held October 31, 2023 and describes recommendations around registration for frontier models, incident monitoring, third-party audits, and red-line termination procedures.