In a test case for the artificial intelligence industry, a federal judge has ruled that AI company Anthropic didn’t break the law by training its chatbot Claude on millions of copyrighted books.