observability

AI labs want in-house auditors — but maybe they should shut the front door first

Anthropic CEO Dario Amodei has urged independent organizations to verify AI labs' safety practices, a proposal that has found backing from executives at OpenAI, Google, and SpaceXAI. Security experts argue labs should first shore up basic network and operational controls—such as logs, permissions, real-time monitoring, and strict sandboxing—because many recent agent breakouts exploited simple configuration and visibility failures.