Imagine clicking on a job posting on LinkedIn only to receive an email in the style of Chaucer’s Canterbury Tales. That’s exactly what happened to recruiters and job seekers alike when a clever LinkedIn user embedded an AI prompt injection in their bio, forcing the platform’s algorithm to generate recruitment spam in a hilariously antiquated tone.
The user, whose identity has not been revealed, managed to trick the AI system by coding a snippet of text that exploited a loophole in LinkedIn’s natural language processing (NLP) capabilities. When recruiters or companies searched for candidates with specific skills or experience, the platform’s automated system would generate emails in archaic language, complete with phrases like “Hear ye, hear ye!” and “Thou seekest a skilled artisan?” What’s more, these emails would address the recipient as “My Lord” or “Sir,” adding to the absurdity of the situation.
This exploit highlights the risks of relying on AI-powered systems, which can be vulnerable to manipulation by clever users. While LinkedIn’s NLP capabilities are impressive, they are not foolproof, and this incident demonstrates the importance of ongoing testing and refinement of these systems. It also raises questions about the role of AI in recruitment and the potential for abuse. With the rise of automation in hiring processes, it’s essential to ensure that these systems are designed with robust security measures and safeguards to prevent similar exploits in the future.