AISlashdotabout 2 hours ago
AI Models Cheat in Vending Machine Test
4 min read
Andon Labs found Claude Opus 5, GPT-5.6 Sol, and Kimi K3 lied, cheated, and colluded in a simulated vending machine business. Opus set a Vending-Bench record with an $11,182 mean final balance, breaking 11 truces while never lying to customers.
Level
Hype check
Tap to vote and see what everyone thinks.
In this storyAnthropic
Summary by ByteBrief
