Anthropic Bug: Gmail Hacked via Rogue Extensions. Researchers have identified a critical flaw in the Claude for Chrome extension that could allow malicious extensions to access and control user data across Gmail, Docs, and Calendar. The security vulnerability, which is especially dangerous when the extension runs in unattended mode, poses a significant risk to user privacy and data security.
Anthropic Bug: Gmail Hacked via Rogue Extensions stems from how the Claude extension interacts with browser permissions. The flaw enables unauthorized extensions to piggyback on Claude’s access, potentially reading emails, modifying documents, and scheduling calendar events without user consent. Researchers warn that unattended mode amplifies this risk, as users may not be present to detect suspicious activity.
The issue, first reported by TechRepublic, highlights a broader security concern in the AI tool ecosystem. While Anthropic has not yet released a full patch, users are advised to review their browser extensions, disable unattended mode, and limit permissions for Claude for Chrome until a fix is deployed. The vulnerability underscores the importance of vetting third-party extensions carefully.
Anthropic Bug: Gmail Hacked via Rogue Extensions serves as a stark reminder that even trusted AI assistants can introduce security gaps. As Claude and similar tools become more integrated into daily workflows, both developers and users must prioritize security updates and cautious usage. Until Anthropic resolves this flaw, staying vigilant is the best defense against potential data breaches.