In 'Claude Fable 5,' conversation history may be read by Anthropic employees; Microsoft is suspending employee use for risk assessment.



Anthropic's AI model 'Claude Mythos,' which was previously only available to a limited number of organizations due to its potential for sophisticated cyberattacks, has been made publicly available as 'Claude Fable 5' on June 9, 2026. It has come to light that a Microsoft employee testified that 'use of this model is not permitted at Microsoft until its data retention policies are evaluated.'

Microsoft restricts Claude Fable for employees over data retention concerns | The Verge

https://www.theverge.com/report/947575/microsoft-claude-fable-5-restricted-internally



The models announced on June 9th are 'Claude Mythos 5' and 'Claude Fable 5.' Claude Mythos 5 is a security-removed version that will be released exclusively to certain organizations, while Claude Fable 5 is a retail version with security measures in place that will be released to the public. Both models are positioned as updated versions of '

Claude Mythos Preview ,' which was announced on April 7th, 2026.

The official version of 'Claude Mythos' has finally been released, and 'Claude Fable,' a version with no user restrictions, has also been released, making it available to everyone - GIGAZINE



Microsoft has made Claude Fable 5 available to GitHub Copilot and Foundry customers. However, it seems that this model is not appearing in the model selection screen of the internal version of GitHub Copilot used by Microsoft employees.

According to a Microsoft employee who spoke to the technology media outlet The Verge, employees have been told that 'the legal team is evaluating Anthropic's data retention requirements, and it is unclear whether they will approve its use in the future.'

With the release of 'Claude Fable 5,' Anthropic also announced that it will implement data retention and review as part of its security measures. Anthropic will retain conversations between users and AI models for 30 days, and a limited number of reviewers will be able to access these conversations 'if a potential for significant harm is identified or if there is a written request from the customer.'

Data retention practices for Mythos-class models | Claude Help Center
https://support.claude.com/en/articles/15425996-data-retention-practices-for-mythos-class-models



According to an employee, an evaluation is underway to determine whether this requirement aligns with Microsoft's internal requirements. Microsoft declined to comment to The Verge.

in AI, Posted by log1p_kr