Pentagon has given Anthropic an ultimatum to relax safeguards on its Claude AI model or face potential penalties, intensifying debate over ethics and military use of artificial intelligence.