DeepFest builds its 2026 programme around live demonstrations

DeepFest is running its 2026 programme around things happening in front of an audience rather than being described to one. Organisers say the four days in Riyadh include real-time deepfake detection and AI-generated music performed on stage, alongside technical sessions, sector talks and training, with more than 180 speakers on the schedule. The choice of those two demonstrations is not arbitrary. They sit at opposite ends of the same technology and make the awkward point that a synthetic-media conference usually avoids: the machinery that composes a piece of music on stage is closely related to the machinery that fabricates a video of someone saying something they never said. Running both in the same programme is a more honest framing than treating generation as the creative story and detection as the security one. Real-time detection is also the harder of the two to demonstrate credibly. Post-hoc forensic analysis of a suspect video can take its time, run several models and produce a probability. Doing it live, on a stream, with an audience watching, removes the margin. Detection accuracy has been the weaker half of the deepfake debate for years, with detectors trained on one generation of synthesis models degrading against the next, and the honest position is that this is an arms race rather than a solved problem. A live demonstration at least shows what the current state of it looks like. The programme sits inside a wider event framing. DeepFest is organised by Tahaluf and run with the Saudi Data and Artificial Intelligence Authority, with HUMAIN as global AI partner, and its published scope runs from AI governance, sovereign infrastructure and autonomous systems through to applications in healthcare, supply chains, media and the creator economy. The event opened this week with speakers drawn from 17 countries , and organisers expect more than 120 AI companies on the exhibition floor. Creator-economy content has become a thread across the whole show rather than a DeepFest speciality. LEAP has built a creator studio into its own floor , and Adobe used the event to open free access to its generative tools to more than 27 million Saudi residents . Generative media is no longer a side track in Riyadh; it is one of the main ones. What organisers have not published is the detail that would let anyone judge the demonstrations before seeing them: which companies are running the detection demo, what models or systems it is built on, what material it is tested against, or how the AI-generated music performance is produced and by whom. Those are the specifics that separate a working system from a rehearsed set piece, and none of them has been released. Conference demonstrations are chosen to succeed, which is worth remembering while watching one. The interesting question about the detection demo is not whether it catches the fakes prepared for the stage. It is what it does with a fake it has not seen before. TechScoop is reporting from LEAP 2026 in Riyadh.