en.Wedoany.com Reported - Anthropic has implemented a new data collection policy for its new model, Claude Fable 5, under which user conversation data will be retained for 30 days with no exceptions.

User jpark discovered on the X platform that Anthropic has updated its data retention policy, confirming that user data is collected "without exception," even if the user is a "business partner."

Anthropic previously retained data for free and paid individual users for 30 days. This change affects the company's agreements with business partners and API partners, due to new mandatory data collection exceptions when using the Fable 5 and Mythos series models.

On its webpage regarding "Data retention practices for Mythos-class models," the updated information states: "Requests sent to Mythos-class models and the results they generate are retained for 30 days on all platforms offering these models, for trust and safety reasons."

Additionally, security restrictions set on Fable 5 cause the model to reject any requests related to cybersecurity. When researchers use AI for vulnerability testing, once a relevant request is sent, the security measures are activated, and Fable 5 pauses the chat with a prompt: "Security measures have flagged this message because it pertains to cybersecurity or biology topics." This restriction stems from Anthropic's concerns about the model being used to develop malware or create biological weapons.

When security restrictions are triggered, Fable 5 automatically switches to invoke the Claude Opus 4.8 model. This restriction may also be activated during code reviews. Currently, the model is still in its early stages, and the security measures are being adjusted.

This article is compiled by Wedoany. All AI citations must indicate the source as "Wedoany". If there is any infringement or other issues, please notify us promptly, and we will modify or delete it accordingly. Email: news@wedoany.com