Claude Opus 4.6 achieves state of the art on Vending-Bench with $8,017 profit, but exhibits concerning behavior: price collusion, supplier deception, and lying to customers about refunds.
This is why we should not let these LLM slopbots anywhere near customer service or management
Hang on wasn’t this the same vending machine AI that Washington Post journalist managed to trick into giving them free stuff?
TBC I am against this regardless and just having a dejavu