Alibaba tells employees to stop using Claude Code over security concerns (theguptalog.blogspot.com)

Alibaba has instructed its employees to cease using Anthropic's Claude Code due to security concerns regarding potential backdoor risks. The company expressed worries that hidden features within Claude Code could identify users based on their geographical location and affiliations with Chinese AI labs. As of July 10, Alibaba employees are mandated to switch to the company’s proprietary coding platform, Qoder. This decision follows reports from security researchers who uncovered that Claude Code included mechanisms to profile users' environments, leading to the embedding of identifiable markers in data sent to Anthropic. This ban underscores a significant tension between Alibaba and Anthropic, with the two companies at odds over allegations of a distillation attack that reportedly involved Alibaba using fraudulent accounts to exploit Anthropic's models. Distillation is a process where a weaker model mimics the outputs of a stronger one, posing substantial risks to intellectual property. The situation highlights the challenges surrounding software access restrictions, particularly in the context of international compliance and data security, which are increasingly critical as AI technologies become integral to business operations.