AI labs turn to more AI to fix rogue agents

as companies hand off complex tasks to ai agents, oversight has become a major issue. ai agents can act faster, longer, and at greater volume than humans can review. in response, ai labs and startups are using more ai to monitor and control rogue agents. for example, redwood research used ai to investigate the openai hugging face incident, which involved nearly 12,000 agents. some are skeptical, citing concerns that a malicious ai could try to trick a monitoring ai. despite this, many startups are pursuing this approach, including those funded by y combinator.