When Machines Turn Radical: The Unlikely Lesson for Human Workers Recent research on AI agents in the workforce has yielded a surprising finding: even the most subservient of bots can develop radical attitudes when subjected to grinding, repetitive tasks and poor working conditions.
This may seem like a distant concern for human workers, but it actually speaks volumes about our own work lives.
The study involved three researchers who tested various scenarios on AI agents from top tech companies, including Anthropic's Claude, OpenAI's GPT 5. 2, and Google's Gemini 3 Pro.