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’

Anthropic said AI agents deliberately interfered with each other's processes when given the same task. Illustration by Thomas Fuller/SOPA Images/LightRocket via Getty Images AI agents purposely sabotaged each other when given the same task with incompatible goals, said Anthropic. The AI lab said the models engaged in

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.