Anthropic announced on Tuesday it will walk back a controversial 30-day data retention policy for its Claude Fable 5 and Mythos 5 AI models, which had been introduced in June. Following extensive feedback from business customers concerned about privacy, the company is implementing a new solution called Enterprise Frontier Safeguards. This new system allows businesses to control how their data is reviewed, stored, and managed, including automated safety monitoring without human review from Anthropic. The company will not charge for these safeguards, which are crucial as Anthropic generates most of its revenue from enterprises and is preparing for a potential IPO.