“xAI has implemented safeguards in its models to prevent them from providing guidance on dangerous virology methods.”