The AI Safety Letter.

We are not anti-technology. We are anti-surrender.

Artificial intelligence is reshaping work, evidence, education, war, and democracy faster than any institution can govern it. No single company, party, or country should decide what kind of intelligence is unleashed on the world, or under what conditions.

We call on our representatives, at every level of government, to publicly support a global AI safety framework before AGI arrives, built on five principles:

  1. Safety: independent evaluation of frontier systems before deployment.
  2. Sovereignty: democratic accountability over civilization-scale AI.
  3. Transparency: legible capabilities, provenance, and deployment scale.
  4. Accountability: real liability for foreseeable AI harms.
  5. Human Primacy: human judgment final over rights, safety, war, and the conditions of human life.

We do not need to choose between progress and survival. We need to choose to govern what we build.