Definition
A 2023 agreement where 28 countries and the EU pledged to cooperate on the safety risks of the most powerful AI systems.
At a glance
- Signed November 1, 2023 at the UK AI Safety Summit (Bletchley Park) — the first global summit of its kind[2].
- Endorsed by 28 countries plus the EU, including the US, UK, and China — an unusually broad alliance[3].
- Non-binding: it sets shared intent on frontier (most-capable) AI, not enforceable law.
What it says
AI should be safe, human-centric, trustworthy, and responsible[1]. It flags frontier-AI risks like cyberattacks, biotech misuse, and deceptive content, and commits signatories to study those risks together as the technology advances[4].
What it means for you
No obligations land on your business directly. But it signals where regulation is heading — toward safe, transparent, accountable AI. Building responsible AI use in early pays off as rules tighten.
Bottom line
A milestone of intent, not enforcement: 28 countries and the EU agreeing that powerful AI carries shared risks worth tackling together.