Anthropic bars Chinese-owned groups as US AI tensions continue to rise

Anthropic extends restrictions to ownership and control

Anthropic updated its terms to bar companies and organizations that are majority owned or controlled by entities in jurisdictions it considers restricted, including China, from accessing its AI services, regardless of where the entities operate.

The company said certain organizations had been routing access through subsidiaries incorporated abroad; the new rule closes that ownership‑based loophole.

This builds on existing restrictions against many organizations in sanctioned or authoritarian regions (such as Russia, Iran, and North Korea) and now explicitly includes China under the ownership control criterion, targeting entities more than 50 percent owned directly or indirectly.

The change zeroes in on subsidiaries and shell structures

In its announcement, Anthropic said organizations from restricted regions were accessing services via subsidiaries incorporated abroad.

The revised policy blocks that practice by tying eligibility to ultimate ownership rather than incorporation locale.

Practically, a Singapore or EU subsidiary that is majority owned by a Chinese parent would now be ineligible, even if it previously passed basic geo checks. The company framed the move as closing risk pathways tied to adversarial state objectives.

Legal and security risks are the cited drivers

Anthropic’s rationale cites certain jurisdictions’ legal, regulatory, and national security risks. The firm argues that capabilities accessed through foreign subsidiaries could be repurposed to serve adversarial military and intelligence ends, countering its safety commitments.

This makes the ownership test a defensive measure rather than a simple geofencing tweak. The company positioned the action as aligned with responsible development and guardrails around dual-use risks.

Analysts call it a first among major US AI firms

Industry watchers flagged the move as a leading US AI company’s first formal, public prohibition of this kind.

While immediate commercial impact may be modest due to existing market barriers, experts say it sets a precedent and could spur peers to consider similar ownership-based rules.

The policy shift, therefore, functions as a signal to regulators and rivals as much as a practical access control.

The policy sits within a broader geopolitical context

US AI providers already restrict direct access in China, with OpenAI’s services likewise unavailable and Chinese champions like Alibaba and Baidu filling the gap.

By extending restrictions to ownership, Anthropic moves beyond location controls amid intensifying US-China tech rivalry.

Government scrutiny of AI supply chains and model misuse has grown, and firms are increasingly expected to align with national security guidance when setting customer eligibility.

China’s response emphasizes the depoliticization of tech

When asked, a Chinese foreign ministry spokesperson criticized the politicization and weaponization of technology and trade, suggesting such practices harm all parties.

That stance mirrors prior objections to export controls and platform restrictions. Practically, however, Chinese users have long relied on domestic models or workarounds like VPNs for US AI tools, limiting immediate operational fallout even as policy symbolism looms large.

Customer growth gives Anthropic room to maneuver

Anthropic recently closed a $13 billion round valuing the company at about $183 billion, and disclosed more than 300,000 business customers with a sevenfold increase in large accounts year over year.

With a swelling enterprise base and multi-billion-dollar run-rate revenue, the company can absorb selective off-boarding without derailing growth plans.

The timing suggests confidence that policy tightening won’t materially slow adoption among approved segments.

The update aligns with US policy momentum

The ownership-based restriction echoes a wider US posture of tightening flows of sensitive technology and know-how to strategic competitors.

While Anthropic’s rule is private, it likely anticipates evolving regulatory expectations and reduces enforcement friction.

For buyers, it clarifies eligibility early in the sales cycle; for Anthropic, it standardizes internal reviews across global teams, lowering compliance ambiguity as customer volumes scale.

Enforcement will hinge on due diligence quality

Blocking by ownership means robust know-your-customer processes are essential. Anthropic must verify beneficial ownership, control structures, and indirect stakes, including layered holdings or trusts.

That raises practical challenges similar to export-control screening, where false negatives and complex corporate trees can obscure absolute control.

Expect expanded screening vendors, contractual reps, warranties, and trigger clauses to unwind service if hidden ownership surfaces later.

The bar targets the majority control while leaving edge cases

Anthropic’s threshold focuses on entities more than 50 percent owned by unsupported-region firms. Minority investments or complicated consortia below that line could be more complex to adjudicate.

The company can still apply discretionary review, but the bright line clarifies most cases. Over time, norms may shift toward cumulative influence tests, mirroring foreign investment reviews that look beyond simple equity tallies.

Competitors may face pressure to follow suit

If one leading vendor hardens its stance, enterprise compliance teams may push others to match. Ownership-based screening could become a de facto industry practice, particularly for frontier-model access and agentic tooling.

Providers that do not align risk become the “path of least resistance,” inviting stricter oversight or reputational risk.

Conversely, aligned policies could reduce regulatory arbitrage and standardize procurement expectations globally.

The market impact inside China remains limited for now

US chatbots and developer APIs are already unavailable inside China, fostering a vibrant domestic ecosystem. The new rule closes that route for Chinese companies that previously used subsidiaries abroad to access Anthropie.

But many had already pivoted to local models due to latency, compliance, and cost. As a result, near-term demand may consolidate around Chinese providers while global partners tighten contractual flows.

The move lands amid record fundraising momentum

Announcing stricter access rights after a $13 billion fundraise at a $183 billion valuation signals confidence with investors and customers.

Markets have rewarded growth-with-governance narratives across AI infrastructure, and Anthropic’s run-rate revenue expansion suggests enterprise demand remains robust even as policies tighten.

In other words, the company chooses selectivity from a position of strength, not scarcity.

Chinese advances add competitive subtext

DeepSeek’s rise with a lower-cost model challenged assumptions of a wide US lead in generative AI. Against that backdrop, Anthropic’s restrictions read partially as competitive containment.

It also ensures its highest-end capabilities do not indirectly buttress rivals’ ecosystems via overseas structures. Whether symbolic or substantive, the message is clear about who the preferred customers are as model capabilities race forward.

Explore how China is stockpiling NVIDIA chips to fuel its massive AI ambitions despite US bans.

The bottom line is a tighter but more precise access regime

Anthropic’s ownership-based ban tightens the gate on who can use Claude and related services, prioritizing legal alignment and security over edge-case revenue.

The company argues this wards off misuse while offering clear rules of the road for legitimate customers.

In a year defined by AI geopolitics and hyperscale funding, the move crystallizes how policy and product strategy travel together for frontier AI firms.

See how US authorities are responding to the alleged smuggling of Nvidia’s powerful AI chips to China.

What do you think about Anthropic aiming to lock down the AI chips because of the US government orders? Please share your thoughts and drop a comment.

Read More From This Brand:

• New Malaysia rules target US AI chip imports
• AMD’s CEO sees AI chip demand soaring past $500 billion soon
• Nvidia boss claims China’s military won’t adopt American AI chips

Don’t forget to follow us for more exclusive content on MSN.

If you liked this story, you’ll LOVE our FREE emails. Join today and be the first to get stories like this one.

This slideshow was made with AI assistance and human editing.