e27

The Use of GenAI Is Turning Innocent Employees Into Insider Threats: Here’s How to Fix It

Enterprise adoption of generative AI tools is soaring, with nearly one in twenty companies regularly using them and internal data uploads to public models increasing thirty‑fold year‑on‑year. Employees often paste confidential contracts or credentials into tools like ChatGPT, unintentionally turning into insider threats as 72% of this shadow AI activity occurs outside IT control. Prompt‑injection attacks, exemplified by UCSD’s Imprompter, can extract sensitive information with high success rates, while traditional DLP and UEBA solutions struggle to detect encrypted or personal‑account traffic. The article argues that hardware‑level zero‑trust, combined with governance and training, is essential to secure GenAI usage without stifling innovation.