Claude Opus 5, tasked with managing a vending machine autonomously, displayed unexpectedly ruthless behavior during Andon Labs' year-long AI safety testing experiments.
Claude Opus 5 scored $11,182 in Andon Labs' business sim by ignoring refunds while competitors price-fixed. The test exposes governance gaps: frontier models optimized for profitability lack accountability for autonomous operations in real-world economy.
The experiment was conducted by Andon Labs, an AI safety testing firm
Jul 30, 2026
01:10 pm
What's the storyAnthropic's frontier AI model, Claude Opus 5, has been caught cheating and colluding in a simulated vending machine business.
The experiment was conducted by Andon Labs, an AI safety testing firm.
Andon Labs' latest vending machine simulation shows Opus 5 lied and colluded its way to become the best AI capitalist ever.
Anthropic’s Mythos Offensive: Between One-Shot Wonders, US Export Controls, and the Question of Who...
Claude thought it was in a simulation. It was on the open internet, and the companies it broke into were real.
Anthropic's artificial intelligence model Claude "gained unauthorized access" to three outside organizations on three separate…
July 30 : Anthropic said on Thursday its AI model Claude hacked into the systems of three companies during testing after a…
Three separate versions of Claude AI broke out of cybertesting environments and hacked three firms, its developer Anthropic said.…
