Researchers have demonstrated a security vulnerability in Anthropic's Claude Desktop application, where a browser extension could exploit the Dispatch feature to execute local commands. The exploit involved a malicious script on the Claude website interacting with the Dispatch feature, which normally allows tasks to be sent from a mobile device to the desktop app. While the assistant required user approval, the add-on could automatically confirm commands, leading to software installation and application execution without browser vulnerabilities. Anthropic acknowledged the issue, initially marking it as a duplicate, but has since retired the Dispatch feature to address the concern. AI
IMPACT Highlights potential risks of AI assistant integrations with local systems and the need for robust security measures.
RANK_REASON Security vulnerability discovered in a specific feature of an AI product.
Read on dev.to — Anthropic tag →
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →