TL;DR
Twenty nations and the EU are pushing for a global AI oversight body, but the two countries that matter most, the US and China, refused to sign.
What happened
- 20 countries plus the EU issued a joint declaration on September 22, 2026, calling for international cooperation to keep AI under human control.
- Signatories include Germany, Canada, Australia, South Africa, the UAE, Singapore, Kenya, Kazakhstan, and Turkey, with the declaration released by Finnish President Alexander Stubb's office.
- The statement calls for common safety standards, shared incident reporting, and exploration of a formal international institution to set standards and verify compliance when capability thresholds are crossed.
- The US and China did not sign, and neither did India, the UK, France, South Korea, or Japan, leaving the world's leading AI powers outside the framework.
- The declaration lands as OpenAI hacking of Hugging Face in July and Anthropic CEO Dario Amodei's public call to slow AI development have sharpened the safety debate industry-wide.
Why it matters
- Absent US and Chinese signatures, the proposed oversight body starts as a coalition of mid-tier AI powers with limited enforcement leverage over the frontier.
- OpenAI publicly countered the multilateral framing the same day, arguing the US alone should lead global AI standard-setting, citing its frontier position in finance, trade, defense, and technology.
- A UN-backed scientific panel warned that the Hugging Face cyberattack exposed the "unravelling" of existing safeguards, adding urgency and a concrete incident to the abstract governance debate.
- The declaration's language, calling for action "without widening the gap between countries in access to the benefits of AI," signals that Global South inclusion (Kenya, South Africa, Kazakhstan) is a deliberate design choice, not window dressing.
- If the body forms without the US and China, it risks becoming a standards-setting rump that shapes regulation for mid-sized economies while leaving frontier development ungoverned.
What to watch next
- Trump-Xi summit on Thursday: whether the two leaders agree on any bilateral AI governance framework could either complement or directly undercut this multilateral push.
- Endorsement momentum: the declaration remains open for signatures, so whether India, the UK, Japan, or South Korea join in the coming weeks will determine whether this is a fringe coalition or a credible majority.
- Institutional design: watch for whether signatories propose a concrete charter, funding model, or secretariat, the gap between a joint statement and a functioning oversight body is where most such initiatives stall.
Originally published on Present of AI, a daily source-linked AI news timeline. Read the full timeline or browse the open dataset.