A new study published in Nature reveals that artificial intelligence models can now generate highly realistic synthetic images that are indistinguishable from real photographs to the human eye. The research, conducted by a team at Stanford University, demonstrates that the latest diffusion models achieve a 99.7% success rate in fooling human evaluators in controlled tests. This advancement raises significant questions about the future of digital media authenticity, the potential for misuse in creating disinformation, and the urgent need for reliable detection tools. The researchers emphasize that while the technology has creative applications, its societal implications require proactive policy discussion and the development of robust verification standards. Read the full article at https://technologyreview.com/2024/05/15/ai-image-realism-study.