Anthropic has been conducting tests to identify issues in how AI agents interact with each other.

The post Conflicting Test Goals Pushed Claude Agents to Deploy Self-Replicating Malware appeared first on SecurityWeek.