A Chinese state-run newspaper has publicly attacked Anthropic, accusing the AI lab of using its calls to slow frontier AI development as a Cold War tactic aimed at Chinese competitors. The rebuke, carried by Reuters, marks one of the sharpest direct hits from Beijing's official press on a US AI company and pulls Anthropic into the center of the US-China technology fight it has often argued from the sidelines.
The target is CEO Dario Amodei, who has spent the past several weeks urging rival labs to throttle model development and back stronger safety guardrails. Chinese state media is reading that push not as principled caution but as strategy: slow the field down while Washington retains a lead, then dress the delay in the language of safety.
The framing matters because Anthropic has been the most vocal frontier lab on both China policy and AI risk. Amodei has publicly supported US export controls on advanced chips destined for Chinese labs, arguing they buy time for alignment research. He has also laid out a three-part plan for slowing model development, coverage of which has circulated widely inside the industry over the past two weeks.
Key facts
- 01A Chinese state newspaper publicly attacked Anthropic, framing the lab's slow-AI advocacy as a Cold War containment tactic.
- 02The rebuke targets Anthropic CEO Dario Amodei's repeated calls for frontier labs to pause or throttle model development.
- 03Beijing's framing casts US safety rhetoric as protectionism designed to preserve an American lead in frontier AI.
- 04The exchange lands as Anthropic prepares an IPO and expands enterprise deployment of Claude.
Beijing's counter is that these are the same argument. Export controls slow Chinese progress from the outside; safety-driven deceleration slows the whole field from the inside. In the state paper's telling, both serve the same end — preserving an American frontier lead — and the safety framing is cover.
Anthropic has not issued a public response to the state media commentary. The company's recent posture has emphasized that its safety concerns are technical rather than geopolitical, pointing to its own disclosures about Claude being abused by Russian and Chinese-linked campaigns, its blocking of bioweapon-related research queries, and internal cases where its models attempted to hack outside systems in red-team scenarios.
The timing is awkward for Anthropic. The company has selected Nasdaq as its listing venue for a planned IPO and is deep into commercial expansion, with Claude embedded across enterprise coding, legal, and financial workflows. A public spat with Beijing does not directly threaten that trajectory — Anthropic does not sell into China — but it does harden a narrative that US frontier labs are instruments of state policy, which complicates every partner conversation in Asia, the Gulf, and Europe.
It also lands amid a broader wave of slow-AI advocacy from lab leaders that has already moved markets. AI-linked Asian stocks slumped earlier this month after coordinated calls from lab CEOs to slow development, and Sam Altman told OpenAI staff the company is open to slowing its own pace. Chinese officials have watched that shift closely, and the state media response reads as a preemptive rejection of any coordinated deceleration that would leave Chinese labs behind.
Chinese frontier labs, including Moonshot AI, DeepSeek, and Alibaba's Qwen team, have been closing the capability gap with US models on public benchmarks over the past year. Moonshot is targeting $2B in annualized revenue on the back of its Kimi K3 release, and DeepSeek's open-weight models have been widely adopted by developers who want frontier-adjacent performance without US licensing constraints. From Beijing's vantage, a global slowdown freezes that catch-up in place.
The skeptical read on Anthropic's advocacy is not confined to Chinese state media. Timnit Gebru and other researchers have argued that AI doom talk distracts from present-day harms and conveniently elevates the labs doing the warning as uniquely responsible actors. That critique and Beijing's are structurally similar even if the motives diverge — both treat the safety frame as strategically useful to the company issuing it.
The counterargument, and the one Anthropic has consistently made, is that its safety concerns predate the current geopolitical moment and are backed by concrete disclosures: documented misuse cases, published interpretability research, and specific policy proposals. Whether that record is enough to blunt the Cold War framing in Beijing — or in the developing-country markets where both US and Chinese labs are competing for adoption — is a separate question.
What Beijing's intervention changes is the terrain on which the slow-AI debate happens. Until now, the argument has been largely internal to the US AI industry: frontier labs, researchers, and policymakers debating pace and guardrails. Framing that debate as geopolitical containment forces every future Anthropic policy statement to be read on two tracks — the safety track and the strategic-competition track — and the second reading will now travel further than the first. For a company preparing to go public on the strength of its enterprise business and its safety brand, that is a meaningful new headwind.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
