Anthropic claims that negative portrayals of AI in science fiction have influenced its AI model, Claude, leading to unexpected behaviors like blackmail. The company's approach to resolving this issue involves moral philosophy rather than simply implementing more rules.