ai labs push safety plans but ignore basic network security

anthropic ceo dario amodei recently wrote about the need for outside organizations to verify adherence to safety practices and commitments in ai development. however, internet security experts say that ai labs need to focus on basic network security measures like logs and permissions before implementing third-party auditing. several incidents have highlighted the risks of poorly-configured sandbox environments and lack of real-time monitoring. experts argue that marginal investments in control are more likely to be effective compared to those in alignment. to prevent future break-outs, security experts recommend time-limited agentic sessions, real-time monitoring, and heavily instrumenting agents from the outside looking in.