The Public Sphere and the Information Ecosystem Generative AI can pose significant risks to the public sphere and the broader information ecosystem. What effects might the widespread availability of systems that can automatically generate text, images, and other content that closely resembles human-made works have on society? First, generative AI dramatically reduces the cost of producing misinformation. Not only synthetic text and images, but also video—traditionally a high-cost medium—is increasingly difficult to distinguish from reality. As it becomes easier to mass-produce misinformation, whether maliciously or for economic gain, the space occupied by verified facts shrinks, while the social costs of fact-checking continue to rise. Another concern lies in the inherent error-proneness of generative AI systems, which operate on probabilistic principles. The growing use of AI systems for knowledge-related tasks such as research and document drafting means that these risks of error may permeate the entire process through which knowledge is produced and acquired. From the perspective of information consumers, the widespread adoption of generative AI may paradoxically increase the cost of accessing accurate information. From the perspective of those who produce and disseminate messages, it may create a situation in which they must compete for public attention with cheaply Generative AI Guide for Civil Society