Prompt-Injection Bug Hits $4B Agentic AI App 'Manus'

A prompt-injection vulnerability in the AI app Manus, a rapidly growing agentic AI platform, allowed researchers at Salt Labs to execute remote code execution and steal credentials from connected third-party services, including Gmail, Dropbox, and GitHub. Despite the vulnerability's existence for nearly four years, it has largely remained a research concern rather than a widespread attack vector. Salt Labs warns that as agentic AI adoption grows, prompt injection attacks are likely to become increasingly common.