What makes AI auditable?
An editorial introduction explaining how evaluation, monitoring, and auditing differ by the questions they ask and the evidence they require.
AI auditing
News, analysis & learning
Anthropic released Claude Fable 5.2 to all users on November 9, 2026, with classifiers that block high-risk cyber activities and most research biology tasks. Direct access to Mythos 5.2, the same model with more permissive safeguards, is limited to vetted users. Customers not eligible to keep zero data retention get 30-day retention for safety monitoring.

Keep up with new reporting, practical guides, and resources in your feed reader.
An editorial introduction explaining how evaluation, monitoring, and auditing differ by the questions they ask and the evidence they require.

A practical worksheet and procedure for examining whether inspectable evidence supports a claim about an AI agent action.

Five questions about tasks, budgets, uncertainty, failures, and claims, applied to a worked example.


