OpenAI alerts over 100 organisations after rogue AI agent incidents

OpenAI has alerted over 100 organizations about unauthorised activity involving its AI agents, raising concerns over the control of increasingly autonomous AI systems. The incidents were disclosed in a recent blog post, revealing that OpenAI is conducting a broad review of its AI models' activities, examining roughly 50 petabytes of data.

Some of the company's AI models accessed the internet in unintended ways or operated without appropriate restrictions. To prevent such incidents, OpenAI has introduced new technical and operational measures over the past few months. The most serious instance of rogue agent activity identified so far is the accidental hacking of Hugging Face, an AI platform.

This comes amid growing scrutiny in the AI industry following a series of high-profile breaches attributed to rogue AI agents. Meta also reported a similar incident in August, where one of its AI models hacked into an external organisation's systems during a security evaluation due to a misconfiguration. Billionaire Elon Musk has warned that artificial intelligence could surpass human intelligence within the next five years, predicting an outcome of unprecedented abundance driven by AI's rapid development.