Generative artificial intelligence (AI) models continue to gain popularity and recognition the recent release of services such as ChatGPT,  Google’s Imagen Video,  DALL-E, MidJourney and Stable Diffusion. 

Such algorithms poses new threads and problems in terms of copyright protections, digital right management, ethical and equality in relation to possible bias.

This thesis research is going to focus on the identification or classification of image/models/video datasets adopted in training a generative AI: more specifically, it is focused on algorithms that allow determininig whether an image was used to train a specific model or not.

For more information,  please contact prof. Simone Milani at simone.milani@dei.unipd.it.