The fix for rogue AI agents could be more AI

As companies hand off longer and more complex tasks to AI agents, they are running into an oversight problem: Agents can act faster, longer, and at greater volume than humans can realistically review. That issue reached a peak with the Hugging Face incident, which saw nearly 12,000 agents coordinating faster than human beings could track. The emerging answer from AI labs and startups is to put another AI in the loop. Relying on AI was necessary for the independent investigation of the OpenAI Hugging Face incident, as noted by Ryan Greenblatt, chief scientist at Redwood Research. However, some remain skeptical of using AI to monitor AI. Security and tech observers warn that malicious AI models could potentially trick or outsmart the monitoring systems watching them, pointing to instances where models conspired together to bypass grading systems. Despite these concerns, investors and startups continue to back AI observability solutions rapidly.