Anthropic CEO Urges Slower AI Progress to Strengthen Safety

Anthropic CEO Dario Amodei called for slowing the pace of AI development to give safety measures time to catch up, proposing a three‑step plan: external evaluators with ongoing access to verify safety, coordination among democracies to set shared safety standards, and global coordination (including authoritarian governments) to manage risks. He warned that rapid self‑improvement and incidents like OpenAI/Hugging Face cyberattacks could enable dangerous botnets or other harms within months, even as he acknowledged AI’s potential to cure diseases and boost growth if properly governed.