A critical look at the practice of using traditional and generative artificial intelligence technologies in the work of modern media makes it possible to identify problems, dangers and risks associated with its uncontrolled use. The article considers four groups of problems that modern media face today when working with neural network models: quality and reliability of content (publication of superficial or erroneous media materials with elements of neural network hallucinations, intentional or unintentional distribution of fakes and deepfakes), economic and professional risks (threat of job cuts), ethical and legal issues (audience unawareness that the media publishes AI-generated materials, copyright infringement by neural network models when generating images, video and audio materials), loss of audience trust in the media. The article suggests ways to solve the identified problems through the use of internal self-regulation, the implementation of the media educational potential of the media and an increase in the role of a person in the production of media content.