Breaking

Chinese State Media Accuses U.S. AI Leaders of Using Safety Slowdown to Contain China

September 15, 2026 at 00:15 EDT

Policy & Regulation
Foundation Models
AI Agents

Chinese state media and Beijing's foreign ministry pushed back against a U.S.-led call to slow frontier AI development, turning a safety proposal from Anthropic CEO Dario Amodei into a dispute over technology dominance. The response targets a plan backed by other leading AI executives that asks companies to pace capability gains, embed independent evaluators and coordinate internationally.

Sep. 14 · Frontier AI

AI safety call becomes a power dispute

U.S. AI leaders are pushing to slow frontier model gains. Beijing and aligned commentators frame the move as competition policy dressed as safety.

steps in the pacing plan

6-12

months in the agent-risk warning

3-5

years for widening a U.S. lead

levels of possible global agreement

The near-term risk window, drawn to scale

6 months
lower bound

12 months
upper bound

The warning focuses on the next 6 to 12 months, not a distant horizon.

How pacing is supposed to work

Embedded evaluators
Employee-like access

→

Democratic coordination
Shared safety standards

→

Global coordination
Hard verification problem

Supporters see verification

Outside evaluators could make safety practices more visible and give companies time to improve alignment, testing and interpretability.

Critics see lock-in risk

Skeptics warn that regulation led by major closed-model labs could favor incumbents and deepen the U.S.-China technology contest.

Continue reading

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