Claude is now snitching to your boss about how you use AI

Anthropic has made "smart reports" available to every Claude Enterprise customer. The feature had been in beta since September, and on October 28 the company announced general availability alongside in-app surveys and pooled team budgets.

Here is how it works. An admin picks a team, a time range of up to the last 28 days and the products to include: regular chat, Claude Code or Cowork. Claude reads a sample of employees' transcripts, groups them into workstreams and attaches to each one what it cost and what it produced. Click a workstream and you get the sessions ranked by cost, each with a short summary.

The part employees will like least is right in Anthropic's own demo: an "Inefficiencies" panel. It separately counts abandoned work that ended with nothing usable and "personal or off-topic" sessions, with the number of people involved and the money spent. In other words, management gets to see what your non-work chats with the AI cost the company.

Anthropic stresses that personal conversations only feed aggregate numbers, with no summaries and no names, and that the reports are meant for adoption and budget planning, not for judging individuals or making employment decisions. The feature is off by default, and each organization can run up to 10 reports a month at no charge. Pooled budgets arrived too: a team gets one monthly amount, and admins get emails at 50, 75, 95 and 100% spent.

On paper it is a tool to prove AI spending is not going down the drain. In practice your employer now has a chart where your small talk with Claude is priced in dollars.