Anthropic accuses Alibaba of stealing AI via 29 million fake queries
Anthropic accused Alibaba of using 29 million fake queries to steal AI capabilities via "distillation attacks." This threatens US tech leadership, as stolen advances could accelerate China's military
US artificial intelligence firm Anthropic has accused Chinese tech giant Alibaba of running the largest-ever campaign to steal AI capabilities from it
Read Full Story at BBC World News →Why This Matters
This dispute exposes a critical vulnerability in the global AI arms race: the exploitation of cloud-based models through deceptive queries. If unchecked, such "distillation attacks" could erode the U.S. advantage in frontier AI development by enabling foreign actors to bypass costly research investments. The fallout could reshape international tech governance, forcing regulators to confront the blurred line between legitimate data scraping and industrial espionage.
Background Context
Alibaba’s rapid ascent in AI mirrors China’s broader strategy to dominate emerging technologies through targeted investments and aggressive talent acquisition. The company’s cloud division has long been scrutinized for its close ties to China’s military-industrial complex, particularly in dual-use applications like autonomous systems and natural language processing. Meanwhile, Anthropic’s accusation underscores a growing trend where U.S. firms accuse Chinese rivals of circumventing export controls and IP protections under the guise of open research.
What Happens Next
The outcome could hinge on whether U.S. regulators treat this as a cybersecurity breach or a trade dispute—two pathways with vastly different consequences. If enforcement actions target Alibaba directly, Beijing may retaliate by restricting access to critical datasets or hardware, further fracturing the global AI supply chain. Alternatively, a negotiated settlement could set precedents for how "AI distillation" is policed, potentially reshaping how companies like Anthropic structure their API defenses.
Bigger Picture
This incident fits a pattern of escalating AI-related friction between the U.S. and China, where intellectual property theft allegations are increasingly weaponized to justify trade barriers. It also highlights the fragility of open-source AI ecosystems, which face mounting pressure to balance collaboration with safeguards against exploitation. As nations race to deploy AI in defense and surveillance, cases like this may force a reckoning over whether neutral innovation can coexist with great-power competition.
