Claude Opus 5 became downright ruthless when he was tasked with managing a vending machine | TechCrunch0

• AI, ai apps, ai coding, ai coding assistant, ai dictation, AI inference, ai music, AI Pope, AI regulation, ai safety, ai-agents, ai-chatbot, Andon Labs, Artificial Intelligence, artificial-intelligence, ArtificialIntellgence, Big Tech, big-tech, Biotech & Health, brain tech, Business / Artificial Intelligence, Business / Big Tech, Claude, CoECI News, Earth Science Technology Office, Energy & Environment Newsletter, Exclusive, Fintech, Gaming News Roundup, Gear / Gear News and Events, Global, Global Precipitation, global warming, globalstar, gpt-5.6 sol, HealthTech, healthtech startups, Home, House, ip-tech, kimi, News, newsletter, Newsletters, Nexstar Media Wire News, Opinions - Technology, People News, Photovogue / Editorial / News, Politics / Politics News, Property / News, Reviews, Startups, TC
• July 29, 2026

For a year now, AI security testing company Andon Labs has tasked frontier models with various real-world tasks to determine how well they perform as agents operating for long periods without human supervision. On Wednesday, Andon published a new installment about how things are going in its Vending-Bench research, where the lab has cutting-edge models