Anthropic ships Smart Reports so managers can audit Claude chats

Anthropic has shipped Smart Reports, a Claude Enterprise feature that lets administrators generate summaries of how their teams have been using the chatbot. An admin picks a team, a time window of up to 28 days, specific Claude products such as Claude Chat or Claude Code, and custom analysis questions; Claude then reviews a sample of transcripts, groups activity into workstreams, estimates list-price costs, and produces a structured report, as Futurism first reported.

The support documentation pitches it as a view of "how a team uses Claude" to "report on the work getting done." Managers can sort worker output by cost and efficiency metrics.

The privacy settings do work. By default every drilldown shows "User" instead of a name, and an owner has to flip a separate setting before anyone can reveal who ran a session; every unveil is logged. Conversations Claude flags as involving health, union activity, whistleblowing, protected characteristics, political views or personal matters feed only aggregate analysis, without session-level summaries. Anthropic says the feature is "not intended for evaluating individual performance or making employment decisions."

The commercial logic is loud. Corporate clients make up approximately 80 percent of Anthropic's revenue, per Futurism, and the company is building out enterprise features as it readies for a public stock offering.

The caveat sits in the company's own track record. Futurism's Joe Wilkins notes that Anthropic's previous "anonymized" information was readily de-anonymized using standard language models. The opt-in toggle to attribute sessions by name is less contentious than the question of whether the default anonymization will actually hold.