1. News
  2. AI models & agents
  3. Anthropic · Workplace AI
Anthropic · Workplace AI2 min read

Claude now shows bosses which chats were 'personal or off-topic'

Anthropic's new smart reports let managers see how teams use Claude, including chats sorted as personal or off-topic. Here is what that means for work accounts.

By the Dashline deskSource: Mashable
Claude now shows bosses which chats were 'personal or off-topic'
Photo: Anthony O'Neil, CC BY-SA 2.0

Anthropic has added a reporting tool to paid Claude Enterprise plans that shows managers how their teams use Claude. One of the labels it can attach to a chat is "personal or off-topic." The feature is called smart reports, and it is available now. A company can switch it on in its organization's settings.

What it does

A manager picks a team, a date range of up to the last 28 days, and which Claude products to include: chat, Claude Code, Claude Cowork, or the combined chat and Cowork experience. Claude then reads the team's conversation transcripts, groups them into workstreams and types of output, such as technical documents, and tallies how much the team has spent.

The results appear as interactive charts. Managers can click into a workstream to see its sessions, then rank or filter them by cost, product, date or type of output. They can also ask their own questions about usage and cost, and save those questions to reuse. The reports can make recommendations too, such as new rules for a code repository.

The "inefficiencies" label

Smart reports also list obstacles workers ran into and how often they came up. They flag what Anthropic calls "inefficiencies." That includes sessions that did not produce usable output and sessions classified as "personal or off-topic."

Limits and worries

Anthropic says the reports are meant to help companies understand adoption and plan their spending on Claude. In its words, they "aren't designed and should not be used for evaluating individual performance or making employment decisions." The tool also refuses questions on certain subjects:

  • Race, gender identity and religion
  • Political affiliation and immigration status
  • Health, union activity and harassment complaints
  • Workers' personal lives

The concern is real, though. Mashable points to reports of companies using AI in layoff decisions, and a survey finding that one in four U.S. managers do so often or all the time. In July, former Meta employees alleged the company used improperly tested AI tools to rank workers for a mass layoff. Meta denied it and said people made those decisions.

What it costs

Each organization gets 10 smart reports a month for free. After that, an extra fee applies. The material does not say how much the fee is.

What to do

If your employer uses Claude Enterprise, assume your chats on a work account can be read and sorted by a manager's report. Keep personal matters on your own accounts. Business owners should note Anthropic's own warning: the reports are for tracking adoption and spending, not for judging individual people.

The takeaway

If your company runs Claude Enterprise, treat work chats as visible to managers and keep personal matters off work accounts.

More articles

You should really read these.

Next step

Let’s talk