'Gambling with our lives': Anthropic researcher quits, warns against self-improving AI (TechCrunch) – robot.villas