Anthropic Policy Chief Rejects AI Self-Regulation as CEO's 3-Step Slowdown Plan Gains Backing
Updated
Updated · CNBC · Sep 16
Anthropic Policy Chief Rejects AI Self-Regulation as CEO's 3-Step Slowdown Plan Gains Backing
3 articles · Updated · CNBC · Sep 16
Summary
Sarah Heck said AI companies cannot oversee their own safety work through an "honor code," arguing at a Washington summit that Anthropic wants government involved because firms cannot be "checking our own homework."
Heck said Anthropic is already in near-constant contact with policymakers, speaking with the White House daily and working with Congress all year as it pushes to address regulation, safeguards and China-related risks before it is too late.
Her remarks extend a debate ignited days earlier by CEO Dario Amodei, who proposed a 3-step plan on Saturday to deliberately slow model-capability gains without giving up U.S. commercial or strategic advantage.
OpenAI's Sam Altman, Elon Musk and DeepMind's Demis Hassabis backed that approach, while Nvidia's Jensen Huang and Meta's Mark Zuckerberg argued new laws are unnecessary and said companies can manage safety without an industrywide slowdown.