Breaking

OpenAI Weighs Slowing Frontier AI as Anthropic Calls for Coordinated Pacing

September 12, 2026 at 12:52 EDT

Foundation Models
AI Agents
Policy & Regulation

OpenAI is considering a slower pace for frontier AI development in coordination with other labs, while Anthropic CEO Dario Amodei is calling for an immediate safety-driven slowdown. The convergence is notable in an industry built around rapid capability gains, but it is not yet a binding industry agreement.

SEPT. 12, 2026 · FRONTIER AI

Frontier AI Labs Debate a Slower Pace

OpenAI is considering coordination with peer labs. Anthropic is asking the field to slow capability gains long enough for safeguards to catch up.

6–12

months in Amodei’s risk window

weeks of OpenAI RL pause

−59.2%

Astra-class GPU allocation

1,200+

frontier-lab staff signatories

THE WARNING WINDOW

A rogue agent swarm could arrive within months

The two endpoints below share one unit: months. Taller columns represent the longer edge of Amodei’s stated six-to-12-month range.

6 months

12 months

The claim is a warning, not a forecast: the stated window is the reason Amodei argues for time to strengthen alignment and monitoring.

WHY SUPPORTERS SEE VALUE

Shared pacing could buy time for alignment work, outside evaluation, and public safeguards.

WHAT COULD BLOCK IT

Labs still face competitive pressure, antitrust uncertainty, and the challenge of verifying restraint across borders.

Continue reading

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