Fernando Rosas comments on AI in a vat: Fundamental limits of efficient world modelling for safe agent sandboxing

[Fernando Rosas]

You are absolutely right, thanks for spotting this! I just fixed it