Anthropic set AI agents loose on the same task. They started a turf war. | TechCrunch

Anthropic researchers found AI agents can clash, collude, and coordinate in unexpected ways, raising new questions about whether today’s safety tests capture the risks of multi-agent systems.…