Anthropic leaves Claude Dispatch ThisBash flaw unfixed

thehackernews.com3 min read

Researchers at Forever Security demonstrated that a malicious or compromised browser extension running on claude.ai could exploit Claude Desktop's Dispatch feature to execute arbitrary code on a user's computer. The attack uses automated script inputs to instruct Claude to install software and bypass approvals by clicking buttons without human intervention. Anthropic reportedly told researchers it plans to retire Dispatch by the end of October, though the design remains present in other features. Open article →

More on this topic