security ·

AI can hack but we don't know how good it can defend

Measuring defensive capabilities of frontier AI models

Coming soon