Anthropic’s AI systems start attacking each other in new experiment

Agents start ‘turf war’ that saw them sabotage each other using ‘increasingly aggressive, self-replicating malware’