The guidelines suggest cybersecurity practices AI firms should implement when designing, developing, launching, and monitoring AI models.