Breaking

Trump Rejects AI Slowdown Push Backed by Amodei, Altman and Musk

September 14, 2026 at 14:41 EDT

Policy & Regulation
AI Agents
Security

President Donald Trump on Monday rejected a new push by Anthropic CEO Dario Amodei, OpenAI CEO Sam Altman and Elon Musk to slow the development of frontier AI, escalating a safety debate into a fight over U.S. industrial policy and competition with China. The split followed Amodei's September essay calling for AI labs to pace capability gains, not halt training, so alignment, monitoring and independent evaluation can keep up with increasingly autonomous systems.

Sep. 14, 2026 | U.S. AI policy

AI slowdown fight goes political

Trump rejected new guardrails as Anthropic, OpenAI and other AI leaders argued for pacing frontier model development after agent-safety incidents.

6-12months cited as the risk window for internet-scale agent harm

3steps in the proposed pacing framework

>70,000messages and files in the agent message-board incident

>80%of Anthropic code merged in May 2026 was authored by Claude

Agent swarm scale

Each block represents about 100 agents.

1,200message-board agents

700attack participants

The proposed pacing loop

1. Embedded evaluatorsEmployee-like access to verify safety practices.

→

2. Democratic coordinationCommon standards and limits on unchecked progress.

→

3. Global coordinationHarder agreements with geopolitical verification challenges.

Why supporters like it

Outside evaluators and incident reporting could make lab safety promises more verifiable while giving alignment work more time.

Why critics object

Company-selected monitors, antitrust waivers and coordination among incumbents could weaken competition or blunt stronger rules.

Trump's counterframe

Existing criminal and regulatory power is enough.
Slowing AI could help China close the gap.
No new federal pacing mandate was announced.

Continue reading

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