Sept 14 — Foreign media report Nvidia and several U.S. firms have begun limiting
or halting use of Anthropic’s frontier models. One immediate trigger was
Anthropic’s unilateral change to its data-retention terms: mandatory 30-day
retention of user interaction data for safety reviews with no opt-out. Since
Anthropic published its first privacy policy in 2023 it has issued 13 revisions;
a comparison with ISO/IEC 27701 raises concerns about increasing user-privacy
risk. Unlike most major AI firms, which say they will comply with lawful data
requests, Anthropic’s policy additionally allows the company, at its discretion,
to share user data with U.S. intelligence agencies without legal process.
Commentators argue the so-called “technical consensus” among the big U.S. AI
players reflects an industry focus on slowing competitors while accelerating
proprietary data collection, intelligence extraction and rule-setting.
