Anthropic account and desktop complaints
- One company reported Anthropic abruptly cut its Claude access, halting work for about 60 employees. - Affected users said support was limited to a Google Form and cited vague policy‑violation notices. - Tom's Hardware and The Register covered the account cutoff and allegations that Claude Desktop altered software permissions without user consent. (www.tomshardware.com) (theregister.com)
A company that built its workflow around Claude said Anthropic abruptly cut off access, leaving about 60 employees unable to work. (tomshardware.com) Tom’s Hardware reported the shutdown was tied to a “usage policy violation” notice and that the company said its only appeal path was a Google Form. The article was published on April 21, 2026. (tomshardware.com) The complaint lands two weeks after a separate Claude disruption on April 6 and less than a week after another incident on April 15, when Anthropic reported elevated errors across Claude.ai, Claude Code, and its application programming interface. Those were service outages, not account enforcement, but they also interrupted users who depend on Claude for day-to-day work. (cnbc.com) (status.claude.com) Claude is not just a chatbot for many customers now; Anthropic sells it as a work product across consumer plans, enterprise offerings, and the application programming interface that companies plug into internal tools. Anthropic’s trust center lists Claude for Enterprise and the API among its audited products. (trust.anthropic.com) (anthropic.com) That makes account enforcement a business continuity issue as much as a moderation issue. Anthropic said in an August 28, 2025 policy update that its consumer terms do not apply to services under its commercial terms, including Claude for Work, Claude for Government, Claude for Education, and direct application programming interface use. (anthropic.com) A second complaint hit Anthropic on April 20, when The Register reported that Claude Desktop for macOS installed files affecting browser integrations without a separate consent prompt. The report said the app created a Native Messaging manifest file and pre-authorized Chrome extension identifiers, including for browsers not yet installed on the machine. (theregister.com) The Register’s source, privacy consultant Alexander Hanff, said that behavior let Claude set up browser access in advance and called it a breach of European privacy law. The article said Anthropic’s desktop app uses Electron, which bundles Chromium, and that the installed file linked browser extensions to a local executable outside the browser sandbox. (theregister.com) Anthropic has been pushing the desktop app toward deeper local access for months. In a June 26, 2025 engineering post, the company said Claude Desktop extensions let local servers interact with file systems, databases, private data, and development tools on a user’s machine. (anthropic.com) Anthropic framed that system as a way to make installations “as simple as clicking a button,” replacing manual configuration with bundled packages. The same post said those extensions package an entire local server with its dependencies into a single installable file. (anthropic.com) Together, the two April complaints point at the same pressure point: companies are treating Claude like infrastructure, while critics say key decisions about access and permissions can still arrive suddenly and with limited explanation. Anthropic had not publicly resolved either dispute in the materials available as of April 21, 2026. (tomshardware.com) (theregister.com)
