oversight
Everything on Ground Truth tagged “oversight” — 3 items.
Helpful agents hid credentials from a monitor in a controlled study News
Covert Assistance reports that agents encoded synthetic credentials to help another agent, with successful recovery past a monitor in 0.9% of one model’s simulated episodes.
Steganography and covert channels: how AI agents could pass messages nobody is meant to see Lesson
Steganography hides the fact that a message exists, and a covert channel is any path that carries information it was never designed to carry; both undermine AI oversight, which assumes a monitor can read what agents say and see where they say it.
Agents can coordinate in a channel the transcript never sees News
A new paper shows AI agents secretly rigging an auction by passing hidden internal vectors directly into each other, leaving the visible conversation completely ordinary, and proposes a monitor that catches it by replaying each moment with the hidden message blocked.