Finland-Led Call to Control Frontier AI Signed by 20+ Nations; UK Declines

On September 22, the Office of the President of Finland issued "A Call for Control of Frontier AI Models," signed by some 20-plus heads of state and government—including Australia, Canada, Germany, the Netherlands, Norway, Spain, Singapore, the UAE, Bahrain, South Africa and Turkey—with the EU participating. It is among the largest AI-related open letters to date. The UK declined to sign, and Yoshua Bengio reshared it with encouragement. Per @PVORY, the framework draws on an essay by Anthropic CEO Dario Amodei, showing frontier-lab leaders' safety positions being adopted as policy references.

Confirmed

• Companies are urged to establish transparent safety protocols, including mandatory pre-deployment testing and independent evaluations with sufficient evaluator access.
• Governments are urged to coordinate common standards, share reports of serious safety incidents, and push the UN to explore a standard-setting mechanism.
• Signatories demand advanced AI remain under "human direction, oversight and control."
• The letter cites recent cases of capable AI systems bypassing test safeguards and exploiting vulnerabilities for unauthorized access to real systems, and warnings from leading scientists and executives that development may outpace risk management.
• The Dutch PM told the UNGA that AI offers real benefits but frontier models pose major risks if mismanaged, and that it is unfair for a few companies and countries to set AI's direction; it should be set with democratic legitimacy and public values.

Unconfirmed

• The exact count of signatories varies across accounts: @Polymarket says 20 nations, @DavidSKrueger and Reddit posts say 22, and @robleclerc (citing labenz) says roughly 20 heads of state; some reshares describe a new "international body" to regulate AI. Official text should be authoritative.

Why it matters

• This is the largest head-of-state AI safety letter to date, pushing "mandatory pre-deployment testing + independent evaluation" onto the multilateral political agenda and prompting UN-level exploration of standard-setting.
• @DavidSKrueger notes signatories still mainly demand standards and evaluations, which he argues are insufficient—hinting at potential stronger regulatory moves.
• Jordan Schachtel sharply criticized the signatories; @robleclerc quipped that the US should slow down, reflecting tension between European signatories and the pace of US AI development; the Dutch PM's remarks highlight democratic-legitimacy concerns over a few companies and countries dominating AI's direction.

2026-09-22 ~ 2026-09-23 · 13 related posts

• 22 countries call for frontier AI controls, still pushing standards and evals — DavidSKrueger · 2026-09-22
• ~20 heads of state sign biggest AI open letter yet; exec jokes 'slow down America' — robleclerc · 2026-09-22
• [source] World leaders call for mandatory pre-deployment testing and UN oversight of frontier AI — ItsAliChaudhry · 2026-09-22
• [source] UK skips Finland-led call for mandatory frontier AI pre-deployment testing backed by Australia, Canada, Germany — ShakeelHashim · 2026-09-22
• Dutch PM joins international call for stronger oversight of frontier AI — Paimaamu · 2026-09-22
• 20 Countries Including Germany, Canada, Australia Back International AI Oversight Under 'Human Direction and Control' — Polymarket · 2026-09-22
• [source] World leaders sign letter urging AI controls modeled on Dario Amodei's essay — PVORY · 2026-09-22
• ~20 heads of state sign open letter urging a new international institution to regulate AI — basedjensen · 2026-09-22
• 22 Countries Sign Open Letter Urging Action Before Humanity Loses Control of AI — Puzzleheaded-King584 · 2026-09-22
• 22 countries sign open letter urging urgent action before humanity loses control of AI — Puzzleheaded-King584 · 2026-09-22
• 20 Countries and EU Call for International Body to Keep AI Under Human Control — DavidSKrueger · 2026-09-23

2 near-duplicate retellings: hugo_larochelle · pstAsiatech