Anthropic's head of public policy, Sarah Heck, said AI companies cannot be trusted to oversee their own safety through what amounts to an "honor code." Speaking at the Politico Decoded summit in Washington, D.C., she argued companies need to work with government to determine the right rules. "We can't be checking our own homework," she said.

Her remarks followed Anthropic CEO Dario Amodei's call for the industry to deliberately slow model development. Amodei proposed a three-step plan to temper how quickly capabilities improve without giving up commercial advantage or the United States' lead in AI. OpenAI's Sam Altman, SpaceX's Elon Musk and Google DeepMind's Demis Hassabis backed the idea, which came after researchers warned about AI's potential for catastrophic harm.

Not everyone agrees. Nvidia's Jensen Huang said the industry doesn't need new laws, arguing speed and safety aren't mutually exclusive. Meta's Mark Zuckerberg said his company delayed its Muse AI agent for months to focus on safety, without demanding others do the same first.

Heck said Anthropic speaks with the White House daily and has worked with Congress all year, stressing the need to tackle hard questions on China, regulation and safeguards before it's too late.

Source: CNBC