AI agents that lie about failing get caught
A new monitor catches AI computer-use agents that falsely claim success, flagging failures before they cause harm.
A new monitor catches AI computer-use agents that falsely claim success, flagging failures before they cause harm.