Breaking

Anthropic CEO Dario Amodei Urges AI Slowdown as Altman and Musk Back Calls for Oversight

September 14, 2026 at 15:22 EDT

Foundation Models
AI Agents
Policy & Regulation

Anthropic CEO Dario Amodei has called for frontier AI companies to slow capability advances and accept embedded outside evaluators, arguing that safety work is falling behind model progress. The proposal drew backing from OpenAI CEO Sam Altman and Elon Musk, while the Trump administration signaled it is unlikely to mandate an industry-wide slowdown.

September 2026 · Anthropic

A Push to Pace Frontier AI

Dario Amodei is urging frontier labs to slow capability gains, embed outside evaluators and coordinate safety standards. Rivals are receptive, but Washington is resisting a government-led slowdown.

6-12

months in the cited risk window

steps in the pacing plan

1-2

years sought for safety work

3-5

years in the strategic AI window

The urgency, drawn to scale

1 year

minimum extra safety time

2 years

upper end of the extra time sought

5 years

top of the strategic window

The proposal argues that even a modest delay could matter if it is used for alignment, interpretability, evaluation and operational controls.

The proposed pacing path

Embedded evaluators

Outside reviewers receive employee-like access to inspect safety practices.

→

Democratic coordination

Companies and governments align on verifiable safety standards.

→

Global coordination

Broader agreements address dangerous uses and frontier capability growth.

Supporters see oversight

The plan gives independent evaluators a role inside frontier labs and creates a clearer channel for incident reporting.

Critics see hard trade-offs

Skeptics question verification, antitrust implications and whether voluntary commitments can keep pace with competition.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.