On September 19, reports emerged indicating that on September 14, foreign media outlets disclosed that several U.S. firms, including Nvidia, have started to limit or discontinue the utilization of Anthropic's state-of-the-art models. The reason for American companies abandoning this large model based in the U.S. is that Anthropic has unilaterally revised its data retention agreement. This revision mandates the retention of user interaction data for a period of 30 days for security reviews, providing users with no option to decline. This is not the first instance of Anthropic modifying its user data agreement. Since the release of its initial Privacy Policy in 2023, Anthropic has made 13 amendments. When compared to the privacy information management system standard ISO/IEC 27701, it becomes evident that there has been a consistent increase in security risks related to user privacy data. Most mainstream AI companies, both domestically and internationally, declare that they will collaborate with law enforcement agencies in data retrieval in compliance with the law. However, Anthropic's privacy policy states that user data can be shared with U.S. intelligence agencies without undergoing legal proceedings, as long as the company deems it necessary. The decision-making power and standards for this sharing are entirely controlled by the enterprise. When considering the "technological consensus" reached by the three U.S. AI giants from this perspective, it becomes clear that it is an internal matter within the U.S. AI industry. They are striving to decelerate the pace at which others catch up while simultaneously expediting their own data collection, intelligence mining, and the establishment of rules and safety standards.