AI Safety

Anthropic CEO Dario Amodei's new essay on AI exponential policy

Amodei warns that AI capabilities could double every 12 months...

Deep Dive

Dario Amodei's new essay, "Policy on the AI Exponential," has been shared as a link-post on LessWrong.

Key Points
  • Amodei proposes a three-tier regulatory framework based on compute thresholds (10^25 to 10^27 FLOPs)
  • Argues AI capabilities double every 12–18 months, making current governance irrelevant
  • Calls for a new federal AI agency and an international treaty within 2–3 years

Why It Matters

Sets the agenda for AI policy debate; directly impacts how tech leaders and regulators must prepare.

📬 Get the top 10 AI stories daily