Anthropic launches Smart Reports for Claude Enterprise to analyze employee AI chats

Administrators can drill into individual workstreams and view session summaries ranked by cost, including date, product, and output type. A separate toggle controls identity disclosure: with it off, drilldowns show "User" and hide session IDs; with it on, names are hidden by default but can be revealed on demand, with every reveal logged. Certain sensitive categories — personal life, health, political or religious views, protected characteristics, union activity, harassment or whistleblower complaints, privileged legal content, and sentiment about a named person — receive only aggregate treatment and cannot be drilled into at the session level. Anthropic says these protections cannot be disabled and custom questions cannot target restricted topics, though the automated classification has no guaranteed detection rate and no failure-rate data has been published.

Smart Reports is not available in all countries where Anthropic operates, with regional limits likely reflecting varying privacy laws and workplace-monitoring regulations. Anthropic notes that product-level controls do not guarantee legal compliance; employers must still meet local obligations such as employee notice, consent, works-council consultation, and data minimization. Enterprise deployments sit inside a separate legal framework from consumer accounts, and organizations must evaluate deployment appropriateness independently.