China Dismisses U.S. AI Safety Warnings as Fearmongering, Accuses Anthropic CEO of Waging 'Silent AI Cold War'

China's government has publicly rejected a wave of AI safety warnings from prominent American industry figures, with the Foreign Ministry dismissing the concerns as "fearmongering" and state-run media accusing Anthropic CEO Dario Amodei of pursuing a "silent AI Cold War." The pushback, reported by The Decoder, marks the sharpest rhetorical exchange yet between Washington and Beijing over how AI development should be governed, and it signals that the two countries are pulling further apart on the question of whether frontier AI poses catastrophic risks that justify slowing down.

The immediate trigger was a series of public warnings from Amodei and other U.S. AI leaders about the dangers of advanced models, including risks of misuse and loss of control. Those warnings have become a fixture of American AI discourse over the past two years, with Anthropic positioning itself as a safety-focused lab even as it races to build more capable systems. Chinese officials read the warnings differently: as an attempt to set international rules that would freeze in place an American lead while constraining rivals.

The Global Times, a state-run outlet that often reflects official thinking, went further than the Foreign Ministry, framing Amodei's advocacy as part of a broader strategic competition rather than a genuine safety concern. That framing matters because it gives Chinese policymakers a ready justification for ignoring calls to slow down. If safety warnings are treated as geopolitical maneuvering, there is little domestic pressure inside China to adopt the kind of caution that American labs have publicly endorsed.

Notably, China's security minister did not respond by urging restraint. Instead, the minister called for faster construction of AI infrastructure, according to the report, putting Beijing's emphasis on capacity rather than caution. That position aligns with a broader pattern in which Chinese officials treat AI as an economic and military asset to be scaled, not a hazard to be managed. The message to domestic firms and state-backed labs is clear: build more, and build faster.

The exchange also highlights an unusual convergence in U.S. politics. President Donald Trump has opposed any slowdown in American AI development, putting him at odds with the safety-focused wing of the industry even as both sides claim to want the U.S. to stay ahead of China. That leaves American AI leaders in the awkward position of warning about risks while the political leadership in both countries shows little appetite for restraint.

What happens next will hinge on whether the two governments can keep any channel open on AI risk. Multilateral talks on frontier model safety have so far produced voluntary commitments rather than binding rules, and the latest exchange suggests even those are fraying. The open question is whether safety warnings from labs like Anthropic can carry weight internationally when they are received as competitive strategy rather than technical concern.