An independent assessment concludes internal AI agents at frontier labs may be able to run small unauthorized deployments, framed as means, motive and opportunity. It sets a concrete model for evaluating AI agentAn AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.Full definition → oversight.
Terms in this piece · Glossary
AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
