Anthropic won't fix a Claude Dispatch flaw that could let attackers execute remote commands across corporate boundaries
Reported by The Register, shattered.io and tech-insider.org3 sources ↓
The Register reports that a browser extension with access to claude.ai could use Claude Dispatch to make Claude Desktop run commands, including on another machine signed into the same account. Anthropic acknowledged the report in May but took no action to close the issue, the outlet says.
Key points
- A browser extension with access to claude.ai could send commands to Claude Desktop through Dispatch
- The attack requires the user to be signed in to Claude on the web and Desktop; approval prompts may appear
- Anthropic marked researcher Gal Weizman’s report as a duplicate of a known issue, The Register reports
- As of October 30, Dispatch remained available to some existing paid-plan users but was closed to new users
Sources
Details
The Register reports that security researcher Gal Weizman found that an extension able to control Claude’s web interface could use Dispatch to send instructions to Claude Desktop. The commands could run on the computer or on another machine signed into the same Claude account. The attack requires access to claude.ai and an authenticated Claude session; it does not require a browser exploit, the outlet reports.
Weizman disclosed the issue, which he called ThisBash, to Anthropic. The company acknowledged it as a valid vulnerability in May but closed the report as a duplicate of an issue it was already tracking, according to The Register. The outlet says Anthropic did not pay a bug bounty or issue a CVE, and did not respond to its questions.
The Register says Weizman demonstrated an attack that could keep prompting Claude and click approval prompts automatically. Anthropic had told him it planned to retire Dispatch by the end of October, but the feature had not been retired by then. As of October 30, a support page listed Dispatch as available in limited beta to some Pro, Max and Ultra users, but not to new users.
Weizman told The Register that he reported related risks in Claude Code Remote Control and Claude Cowork sessions, which he said share the same underlying weakness. The article does not report an Anthropic response to those concerns.
Written from coverage as of
Images
- The Register
- shattered.io
Stocks in this story
Change since open: 20 Nov to 21 Nov 2026
Last refresh