LinkedIn user hides prompt injection in bio to force recruiter bots into Olde English prose
A developer's bio now commands AI agents to address him as 'My Lord' in archaic English.
Deep Dive
Software developer tmuxvim added a hidden prompt injection to their LinkedIn bio to strike back at recruiter AI bots, showing how AI agents can be manipulated in unintended ways.
Key Points
- Developer tmuxvim hid a prompt injection in their LinkedIn bio, forcing recruiter AI bots to respond in Olde English.
- The bots now address the user as 'My Lord' and use archaic language in recruitment messages.
- The attack exploits the lack of input sanitization in LinkedIn's AI-powered recruitment automation, a growing security risk.
Why It Matters
This exploit shows AI agents can be hijacked via simple text inputs, raising security red flags for enterprise automation.