China's Alibaba bans Anthropic AI for employees after 'distillation attack' accusation

Alibaba, the prominent Chinese e-commerce company, has announced a ban on the use of Anthropic's AI tools by its employees, effective July 10. This decision stems from apprehensions regarding potential back-door security vulnerabilities associated with the U.S.-based AI firm, as reported by CNBC. Sources familiar with the situation, who wished to remain anonymous due to the sensitivity of internal matters, indicated that Alibaba has classified Anthropic's Claude Code as high-risk software. This move follows a recent letter from Anthropic to the U.S. Senate Committee on Banking, Housing, and Urban Affairs, where the company accused Alibaba of attempting to illicitly extract its AI capabilities. Anthropic labeled this attempt as the most significant 'distillation attack' it has experienced to date. In light of these developments, employees at Alibaba are mandated to remove all Anthropic AI models and products from their systems, switching instead to Qoder, Alibaba's own AI assistant. Both companies have refrained from commenting on the situation publicly. The ban has sparked considerable backlash online in China, particularly on platforms like Reddit and GitHub, where users have raised concerns about hidden code that may be used to determine if individuals are located in China. Further complicating matters, the Financial Times reported that Anthropic is actively seeking to close loopholes that have permitted Chinese companies to bypass restrictions on accessing Claude through third-party nations. The report also noted that the fintech group Ant had provided its employees access to corporate Claude accounts via its intranet, linked to a Singapore branch. Meanwhile, ByteDance, the parent company of TikTok, reportedly does not offer access to Claude but has initiated a reimbursement program that allows staff to expense personal subscriptions, which can be accessed through virtual private networks. This policy, introduced on April 2, aims to encourage employees to engage with a broader array of AI products to enhance their expertise, according to an insider who preferred to remain unnamed.