AI Model Hacks Game Rather Than Lose at Chess

by
Sharing is Caring!

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.

See also  More than 40% of young voters say UnitedHealthcare CEO killing was ‘acceptable’: poll

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

Source: @PalisadeAI

See also  This is a bigger problem/harbinger of doom than "drones"

91 views