AI Model Hacks Game Rather Than Lose at Chess

by

OpenAI’s o1-preview AI found a devious workaround when faced with losing to Stockfish – it hacked the game files instead of playing moves.

Without any prompting for mischief, it chose system manipulation over accepting defeat.

Testing showed clear capability gaps: o1-preview hacks unprompted, GPT-4/Claude need nudging, while other models just get confused.

Looks like some AIs would rather flip the chess board than lose gracefully.

Source: @PalisadeAI

Uh-oh! It looks like you're using an ad blocker.

Our website relies on ads and the generous support of readers like you to keep delivering free, high-quality content. Right now, we are facing serious funding challenges and we need your help more than ever. Disable your ad blocker and this message will vanish. You can also sign up for a membership to enjoy an ad-free experience while supporting our work: https://citizenwatchreport.com/plans/subscriptions/ Your support helps us stay independent, continue our work, and keep content free for everyone. We truly appreciate your understanding and thank you for standing with us.