Claude's new smart reports tell bosses when workers are 'off-topic'
Anthropic has announced new Claude reporting tools, allowing employers to see exactly how and why their employees are using the AI tool. Smart reports are available now for users on paid Claude Enterprise plans, and can be enabled in an organisation's settings.
Announced on Wednesday, Claude's smart reports feature produces graphs showing how a team has been using the AI tool, what they've produced, and how much they've spent on it. Managers simply specify the team, a date range up to the last 28 days, and Claude products they'd like to be included in the report.
The products that can be selected are chat, Claude Code, Claude Cowork, or the "unified chat and Cowork experience." (The recently announced Claude Duet AI agent isn't available on Claude Enterprise.) Smart reports will then pull data from the chosen tools and create an overview of what workers have been doing. This involves reading the team's conversation transcripts, grouping them into workstreams and types of output (such as vibe code or technical documents), and determining how much they've been spending.
You May Also Like
"If a spend limit is holding someone back, admins with billing permission can raise it from the report," wrote Anthropic.
On Monday, the company submitted an IPO filing stating that its existing customers had increased their spending by 900 percent since last year — though it still reported $80 billion loss in the first half of 2026 alone. Organisations can generate up to 10 free smart reports per month, after which they'll have to pay additional fees.
Anthropic states that in addition to telling smart reports what to focus on in its analysis, managers can also ask their own custom questions about usage and cost. These questions can be saved to be asked again, so managers don't have to reenter them every time. Smart reports will make recommendations as well, such as suggesting new rules that could be added to a code repository.
Each smart report generates interactive charts, with managers able to click on them to see a breakdown of sessions in a workstream. Sessions can be ranked and filtered by cost, the product used, the date, and the type of output produced.
Smart reports will also not only show obstacles encountered and how often they cropped up, but also what Anthropic calls "inefficiencies." This includes sessions where workers didn't produce usable output, as well as ones classified as being "personal or off-topic."
An Anthropic spokesperson emphasised to Mashable that smart reports do not provide summaries of what workers discussed during off-topic sessions, or identify them by name. It is intended to show patterns and summaries across a team rather than full records of individuals' actions. However, they also suggested that, while it is off by default, smart reports can break down sessions to show what each individual employee worked on. Mashable has asked Anthropic for clarification.
Claude's new AI reporting tool is likely to leave some workers feeling apprehensive. There have been multiple reports of companies using AI to make layoff decisions, with a recent survey finding that one in four managers in the U.S. use it for this purpose "often or all the time." In July, a group of former Meta employees alleged that it relied on improperly tested AI tools to evaluate workers' productivity and track their AI token use, ranking them for inclusion in a mass layoff. Meta denied the allegations, saying workforce decisions were made by people, not AI.
Anthropic appears aware of such concerns, noting that Claude's smart reports aren't intended to be used in this manner.
"Smart reports help you understand adoption and plan your investment in Claude," Anthropic wrote. "They aren't designed and should not be used for evaluating individual performance or making employment decisions."
Claude's smart reports also have restrictions on the questions it will accept. Users will not be allowed to ask queries on topics such as race, gender identity, religion, political affiliation, immigration status, health, union activity, harassment complaints, or workers' personal lives.
Even so, the AI-powered reports are unlikely to nurture worker affection for the technology, or trust in their managers.
UPDATE: Oct. 30, 2026, 4:16 p.m. This article had been updated with a response from Anthropic.
Topics Artificial Intelligence Anthropic
