Comment by anematode
2 days ago
> Do you think there is such a thing as perfect security? No one can "get it right" in the face of arbitrarily high intelligence
Why didn't they run the model against the sandbox first? They have effectively unlimited spend.
That's the alarming thing about this result: they did run the model in the sandbox, in the sense that they believed there was no internet access for the model.
“Against the sandbox” and “on the sandbox” are not the same thing.
You're suggesting that @anematode was asking why they didn't test the sandbox escape first? Yeah, I don't know. I've read other statements by both OpenAI and Anthropic about that very kind of test, so maybe they had, or believed they had, and it hadn't escaped in those tests. The behavior of these systems isn't deterministic, which is part of the problem.