Back to News
PolicyAI Understanding briefing

Anthropic CEO proposes a three-step plan to slow AI development

CNBC reports that Anthropic CEO Dario Amodei urged AI companies to pace improvements in model capabilities while preserving commercial competitiveness and U.S. leadership.

4 min readRead the primary source
Source-page capture accompanying Anthropic CEO proposes a three-step plan to slow AI development
Attributed reportingSource recorded
Publisher
cnbc.com
Source link
cnbc.comhttps://www.cnbc.com/amp/2026/09/12/anthropics-amodei-proposes-plan-to-slow-the-pace-of-advancing-ai-capabilities.html
Source type
Reporting by a news outlet — not a first-party document.
Also cited

What we could not confirm independently: This claim is attributed to the named outlet. We did not verify it against a first-party document. (cnbc.com)

Story last revised

ContextUnderstand this in 60 seconds

Start here

Test yourselfAI Models Explained Quiz

What changed since publication

  1. First published
  2. CNBC’s report materially advances the same continuing event as the eligible canonical update by documenting Amodei’s September 12 essay and its stated aim: a three-step plan to slow AI capability development while preserving commercial advantage and U.S. leadership. The supplied report does not provide the plan’s individual steps or evidence of implementation.

What happened

CNBC reports that Anthropic CEO Dario Amodei published an essay urging AI companies to intentionally slow the pace at which they improve model capabilities. He proposed a three-step plan intended to temper development without sacrificing commercial advantage or the United States’ lead in AI, while acknowledging that some measures may be easier to implement than others. The supplied report does not describe the individual steps.

CNBC reports that Amodei published the essay on Saturday, September 12, urging AI companies to pace how quickly they improve model capabilities. The report describes the proposal as a three-step plan, but it does not reproduce or explain those steps in detail.

According to CNBC, Amodei said the approach could temper development without sacrificing commercial advantage or U.S. leadership in AI. He also acknowledged that some parts of the proposal may be easier to achieve than others.

CNBC places the essay in the context of calls from a growing number of researchers for a coordinated slowdown. It also reports that the essay followed an Anthropic researcher’s departure and public announcement earlier in the week. The supplied source does not independently confirm the essay’s contents beyond CNBC’s account or identify any formal commitments by other companies.

Source details: cnbc.com

Why it matters

The proposal places coordinated limits on frontier AI progress into the public discussion as a matter of industry practice, rather than treating safety only as a question of safeguards around already-deployed systems. If adopted, pacing could affect how quickly companies train, release, and scale increasingly capable models. CNBC’s report does not establish that Anthropic or other companies have agreed to the plan, nor does it show how the proposal would be enforced or reconciled with competitive pressure.

Pacing the improvement of frontier models could become a significant policy and governance question because it concerns the rate of capability development itself. Slower progress might give safety evaluators, policymakers, and the public more time to assess new systems, but the source provides no evidence about the likely effect of Amodei’s proposal.

The commercial tradeoff is central to the proposal. CNBC reports that Amodei framed the plan as compatible with maintaining business advantage and U.S. leadership, but the report does not explain how companies would coordinate without creating incentives for competitors to move faster.

The proposal is consequential as a stated position from the head of a major AI company, but it remains a proposal in the supplied report. There is no documented implementation timeline, enforcement mechanism, agreement among companies, or evidence that model availability or training practices have changed.

What to watch next

The key next development is whether Anthropic provides operational details for the three steps and whether other AI companies, governments, or independent safety researchers respond. Watch for evidence of concrete commitments, shared evaluation standards, release delays, or other measurable changes in development practices. The supplied report also does not establish whether the essay changes Anthropic’s current model plans, commercial strategy, or expected IPO timing.

Anthropic’s next public explanation should clarify what the three steps require, which organizations would participate, and whether the plan depends on voluntary coordination or public policy.

Evidence of implementation would include specific changes to training schedules, model-release decisions, evaluations, reporting, or access for external safety researchers. None of those changes is documented in the supplied CNBC report.

The report mentions Anthropic’s expected IPO preparations, but it does not establish a connection between the proposed slowdown and the company’s financing plans. The timing, commercial consequences, and effect on U.S. competitiveness therefore remain unknown.

The supplied report does not state whether governments, competitors, or researchers have endorsed or rejected the proposal. Their responses will determine whether the essay becomes a broader industry initiative or remains an individual executive position.

Related guides & quizzes

AI Models ExplainedAI EthicsFuture of AITest what you know — try a free AI quizLook up an AI term in our glossary

Updates and corrections

This canonical story is updated in place when the developing event materially changes. Its URL and original publication date never change.

  • CNBC’s report materially advances the same continuing event as the eligible canonical update by documenting Amodei’s September 12 essay and its stated aim: a three-step plan to slow AI capability development while preserving commercial advantage and U.S. leadership. The supplied report does not provide the plan’s individual steps or evidence of implementation.
See the public corrections log
Found this useful?