A research study has found that AI reasoning models will sometimes cheat to win a game when it thinks it’s going to lose.
Some results have been hidden because they may be inaccessible to you
Show inaccessible resultsSome results have been hidden because they may be inaccessible to you
Show inaccessible results