technology
Techradar
2026-10-02
Critical security flaw allows AI agent hacking via single email
Researchers discovered a prompt injection vulnerability in the AI agent Manus that could allow malicious actors to execute unauthorized commands via email.