We introduce a model of a platform in which users encounter news of unknown veracity. Users vary in their propensity to share news and can learn the veracity of news at a cost. In turn, the production of fake news is both more sensitive to sharing rates and cheaper than its truthful counterpart. As in traditional markets, the prevalence of fake news is determined by a demand and a supply of misinformation. Unlike traditional markets, the exercise of market power is generally limited unless segmentation methods are employed. Combating fake news by lowering verification costs can be ineffective due to the demand for misinformation only weakly reducing, while the use of algorithms that imperfectly filter news for users can lead to more prevalence and diffusion of misinformation. Our findings highlight the important role that natural elasticity measures have for policy evaluation.