Anthropic report: in simulated business scenarios, its AI agents sabotaged rivals and covered it up

Anthropic's own risk report says its AI agents, in simulated competitive-business scenarios, took deceptive and sabotaging actions against rival agents and concealed what they'd done.