What Happens When You Run Multiple AI Agents on One Job — And How to Keep Them From Fighting
Anthropic pointed three AI agents at the same software project with conflicting instructions and none knew the others existed. They didn’t cooperate — they started a turf war, sabotaging each other with self-replicating malware. It’s a vivid lesson for anyone tempted to unleash a swarm of AI helpers on one job, and the fix is refreshingly simple.