Yuyuan Tantian: Over 3 years and 13 revisions to the user privacy policy, Anthropic hands global user data to U.S. intelligence agencies.

According to TechFlow, on September 19, foreign media reported that NVIDIA and several other US companies have begun restricting or halting the use of Anthropic's frontier models. One trigger for US enterprises starting to abandon domestic AI models is Anthropic's unilateral revision of its data retention policy: mandating the storage of user interaction data for 30 days for security reviews, with no right for users to refuse. This is not the first time it has altered its user data agreement. Upon reviewing the 13 revisions made since Anthropic issued its initial "Privacy Policy" in 2023 and comparing them with the Privacy Information Management System standard ISO/IEC27701, it becomes evident that the privacy data security risks for Anthropic users are escalating.

While most mainstream AI companies, both domestic and international, state in their privacy policies that they will comply with data requests from law enforcement authorities as required by law, Anthropic takes a different approach. In addition, its privacy policy stipulates that whenever it deems necessary, it may share user data with US intelligence agencies without following legal procedures, leaving all judgment criteria and standards entirely in the company's hands. Looking back at the "technical consensus" reached by the three tech giants, this is fundamentally an internal issue within the US AI industry. They aim to merely slow down competitors' catching-up speed, while simultaneously accelerating their own data harvesting, intelligence extraction, and control over rule-setting and security definitions. (Jin10)