Tim DaviesCEO Projekt Rising - The expert AI partner for SMEs

AI Image Disclaimers: The Future of Content Creation

12 Sept 2025

Watch on YouTube

Short excerpt from a YouTube conversation (recorded Sep 2025) about labeling and authenticating AI-generated media. A speaker proposes mandatory disclaimers on AI images; Tim Davies argues that watermarking/fingerprinting of images, video and text outputs (e.g., Google Gemini, OpenAI approaches) will become necessary and may be required by law to prove authenticity and to control training-usage of synthetic content.

I personally would like, for example, future AI images to have to have a disclaimer on them to say this was AI generated.
I think we'll end up with a necessity that all video and image content is watermarked in some way, shape or form. And currently a lot of the models are fingerprinting behind the scenes, both on image, video and text output. So actually, the Gemini output, Google knows that it's been Produced by Gemini. OpenAI has a similar approach, so they can effectively, if they choose to stop that content from being used for training purposes. So there will be a case of using synthetic data to train these models to be more capable, but they can specifically exclude that, and I think that will become a requirement in many countries to prove authenticity.