Anthropic set AI agents loose on the same task. They started a turf…

TechCrunch reports that 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.

For Australian readers, the useful question is what this changes in everyday work: costs, productivity, privacy, hiring, or the tools businesses rely on. We are watching this story because practical AI adoption is moving quickly, and the impact often reaches workplaces before policy or training catches up.