Anthropic Policy Chief: AI Companies 'Can't Be Checking Their Own Homework,' Calls for Government Oversight
Summary
- • Anthropic's Sarah Heck says AI companies can't self-regulate, must work with government
- • Comments reinforce CEO Amodei's weekend call to deliberately slow AI model development
- • Industry split: Altman, Musk, Hassabis back slowdown; Huang and Zuckerberg resist new rules
- • Anthropic talks to White House daily and has worked with Congress all year on AI policy
Details
Anthropic policy head rejects AI self-regulation 'honor code'
Sarah Heck at Politico Decoded Summit: 'We can't be checking our own homework, and that's very clear' — direct call for external government oversight
Heck calls for government partnership on AI oversight framework
'We want to work with government to figure out what makes sense' — Anthropic talks to White House 'daily basis' and engaged Congress 'all year'
Follows Amodei's three-step proposal to deliberately slow model development
Amodei proposed tempering capability gains without sacrificing commercial or US strategic advantage; Heck's remarks reinforce coordinated Anthropic pro-regulation messaging
Altman, Musk, and Hassabis publicly back Amodei's slowdown proposal
Musk posted 'Dario is right' on X; OpenAI and DeepMind leaders' support reflects a safety-lab consensus on the need for external oversight
Huang (Nvidia) and Zuckerberg (Meta) resist new AI regulation and slowdown
Huang: industry 'doesn't need new laws or regulations'; Zuckerberg: Meta delayed Muse agent 'several months' for safety but did not call for industry-wide action
Meta delayed personal AI agent Muse release by 'several months' for safety review
Zuckerberg said Meta acted unilaterally — 'We just did it as part of our day-to-day work' — rather than advocating for an industry-wide development pause
Heck: China policy, AI regulation, and safeguards must be addressed 'now'
'If we're not doing it now — whether it's how we deal with China, how we do AI regulation — it may be too late'
Source: CNBC Technology, September 16, 2026. Anthropic policy chief Sarah Heck at Politico Decoded Summit, Washington DC.
What This Means
Anthropic's public call for government oversight of AI — coming from its own head of policy — is a strategically significant move that separates the company from industry peers who prefer self-regulation. The timing, days after Amodei's development-slowdown proposal and its endorsement by Altman and Musk, suggests coordinated messaging from safety-oriented labs seeking to set the terms of any regulatory framework before regulators do it for them. The fault line between Anthropic/OpenAI/DeepMind and Nvidia/Meta on regulation mirrors their deeper divide on development pace — with frontier model labs accepting constraints that deployment-focused platforms view as competitive handicaps.