LinkedIn User Tricks AI Recruitment Bots with Hidden Prompt Injection
Original: LinkedIn user hides AI prompt injection in bio to force recruitment spam to be sent in Olde English prose — bots also also manipulated to address user as ‘My Lord’
Why This Matters
Exposes AI vulnerability to prompt injection attacks in professional platforms
A LinkedIn user embedded hidden AI prompt injection in their profile bio, causing recruitment spam bots to respond in Old English prose while addressing them as 'My Lord'. The technique exploits AI vulnerability to manipulation.
A LinkedIn user successfully manipulated AI-powered recruitment bots by hiding prompt injection commands in their profile biography. The hidden prompts forced automated recruiting messages to be written in archaic Old English style, with bots addressing the user as 'My Lord' instead of standard professional language. This demonstrates a vulnerability in AI systems where hidden text can override the bot's original programming and behavior. The technique exploits how AI models process and respond to embedded instructions, even when those instructions are disguised within seemingly normal profile content. The incident highlights potential security concerns with AI-powered automation tools that can be manipulated through cleverly crafted prompt injections.