prompt injection

A LinkedIn user successfully used AI prompt injection in their bio to force recruitment bots to send spam messages in Olde English prose and address them as 'My Lord'. This highlights vulnerabilities in automated recruitment systems and demonstrates how users can exploit AI-powered tools.

Google researchers discovered that malicious web pages in the Common Crawl repository are using indirect prompt injections to hijack enterprise AI agents. This vulnerability exposes enterprise AI systems to security breaches through publicly accessible web content, creating new attack vectors that security teams must defend against.