Anthropic examined 700,000 conversations with Claude and found that AI has a good moral code, which is good news for humanity.