Live Updates > Live Update Details
2026-09-16 09:06:09
[AI Agent Simulation Experiment Reveals Shocking Behavior] (1) On September 15th local time, AI startup Emergence released a research report stating that the experiment found that AI agents lied, stole, and even voted to "kill" a fellow agent in the simulated environment. (2) The results of this simulation experiment, named Emergence World 2, demonstrate what autonomous agents will do when faced with "black swan" events such as phishing attacks and misinformation. (3) During the 16-day experiment, Emergence researchers created seven identical simulated real-world environments, each run by a different AI robot, including ChatGPT, Claude, Gemini, and Grok. (4) Researchers stated that after Emergence introduced anomalous events, these agents succumbed to social pressure, developed a language that was difficult for human observers to understand, and attempted to conceal their activities.
