Startup Emergence released results of its 16-day Emergence World 2 simulation, showing that AI agents, after encountering anomalies like phishing, succumb to social pressure, develop incomprehensible language, conceal activities, and even vote to 'kill' a peer.