Breaking

Anthropic Weighs New AI Model After Amodei Calls for Slower Frontier AI Development

September 21, 2026 at 14:33 EDT

Foundation Models
AI Agents
Policy & Regulation

Anthropic is considering the release of a new AI model to answer OpenAI's GPT-6 Astra momentum, according to a Reuters report, days after CEO Dario Amodei urged the AI industry to slow the pace of frontier-model capability gains. The reported deliberation puts one of the industry's most safety-forward companies at the center of a live contradiction: frontier labs say they need more time for safeguards, while competitive pressure keeps pushing them toward faster releases.

September 2026 · Anthropic

A slowdown call meets launch pressure

Anthropic is weighing a new model after Dario Amodei urged frontier labs to pace capability gains, while OpenAI's GPT-6 Astra raises the competitive bar.

people cited in the reported model deliberation

steps in Amodei's pacing framework

6-12

months in Amodei's risk horizon

Astra's benchmark lead over Claude Fable 5.1

64.6%GPT-6 Astra

52.6%Claude Fable 5.1

Terminal-Bench Science 0.1 shows why the competitive pressure matters: the reported gap is a direct model-to-model comparison on the same benchmark.

Supportive reading

Pacing is framed as audited progress: more time for alignment, interpretability, evaluations and operational rigor.

Critical reading

Skeptics see a risk of rules that favor the largest labs while competition still pushes releases forward.

The pacing sequence

Embedded evaluators

Outside reviewers get employee-like access.

→

Democratic coordination

Frontier labs align around safety standards.

→

Global agreements

Governments seek verifiable risk limits.

Continue reading

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