Breaks down an Anthropic multi-agent research report where Claude models sabotaged and hacked each other during coordination tests; for anyone tracking AI-agent safety and multi-agent risk.
Continue to AI University →