Anthropic CEO urges AI firms to slow development over misuse fears

Anthropic CEO Dario Amodei urged AI companies to slow model capability development, proposing a three-step framework to manage risks of misuse, as fears over AI's potential harm intensify. Amodei's plan includes independent reviewers inside leading AI firms, coordination among frontier companies to set safety standards, and international cooperation. His call follows a threat report detailing Claude AI misuse and a researcher's resignation warning AI could "kill us all." OpenAI's Sam Altman and xAI's Elon Musk voiced agreement. Amodei stressed slowing progress without halting it, while both Anthropic and OpenAI prepare for IPOs, and he advocated preserving the U.S. lead over China through chip controls and collaboration with government.