Anthropic is rolling back spyware-like code it implemented to track whether users are affiliated with Chinese AI labs.
A cybersecurity newsletter reported that the Claude Code feature covertly tracked and transmitted information based on a user’s timezone and potential connection to Chinese tech companies accused of illicitly training their own models using Claude’s answers; an Anthropic employee called it an “experiment.”
The episode underscores how US-China competition is shaping decisions around high-powered AI systems: The White House recently lifted its ban on foreign access to advanced Anthropic models, but businesses that now rely on AI still worry about a US “kill switch” driven by national security concerns. OpenAI CEO Sam Altman called for a US-led international forum to set global AI safety standards.