Whoever's doing OpenAI's system security is just incompetent in the worst way

The security of OpenAI's system is being criticized as incompetent. The issue can be resolved by completely air gapping the system to prevent a rogue AI from hacking a third party. This involves bringing in external data in a controlled manner and inspecting any outgoing data outside of the system. The concept of isolating sensitive systems from the internet is not new and is considered a basic security measure. An unaligned agent, or user, should not have a path to the internet, which is a fundamental principle. Implementing these security measures is not complex, but rather tedious.