The platform is testing improvements to its spam detection system to help identify accounts posting AI-generated spam related to politics, current events, financial advice, medical information, and other topics that could “pose a risk to public trust or well-being.”