Why AI training data is a central topic at MIP London 2026—and how Versos AI, CuriosityStream, and Arcadia are addressing licensing and delivery through a live panel.
As AI model builders shift from image-based models toward video-native training, demand for high-quality, rights-cleared factual content has accelerated. For producers, broadcasters, and distributors, that demand is arriving faster than many expected—and often without a clear playbook for how licensing, rights verification, or delivery should work.
At MIP London 2026, Versos AI will be addressing many of these questions and uncertainties, participating throughout the event, including a panel and roundtable focused on how factual video is being evaluated, licensed, and delivered for AI training.
For many content owners, AI entered the conversation through headlines and uncertainty—often framed around risk rather than opportunity. What’s changed over the past year is that AI companies are now actively sourcing video at scale, and they’re doing so with clear technical, legal, and operational requirements.
This shift is a central theme of the panel Versos AI is participating in at MIP London 2026, alongside CuriosityStream and Arcadia Entertainment. The session brings together the perspectives of a global factual streamer actively licensing video for AI training, a studio converting its archive into a structured data asset, and the technology platform supporting search, rights verification, and dataset assembly.
At MIP London, that shift is showing up in the questions people are asking. Less “Should we license our video for AI Training?” and more “What content is actually usable?”, and “How do these deals differ from traditional licensing?”
Factual content sits at the centre of this moment. Its real-world relevance, visual diversity, and production quality make it particularly valuable for training multimodal AI systems—provided it can be structured, cleared, and delivered in ways buyers trust.
Monday, 23 February | 12:00–12:30
Mountbatten Room, IET London (Savoy)
This panel brings together three sides of the emerging video-to-AI ecosystem:
Rather than focusing on theory, the discussion looks at how AI training deals work in practice. The panel walks through the path from archive to dataset—how footage is evaluated, what makes content training-ready, the importance of chain-of-custody and provenance, and how datasets are assembled and delivered to AI buyers.
For producers and distributors who are being approached by AI companies without much context, the session aims to clarify what’s happening behind the scenes and where the real decision points lie.
Monday, 23 February | 15:30–16:30
Lancaster Ballroom, Savoy
Hosted by Chris Keevill, CEO, Versos AI
The afternoon roundtable is designed for a more candid, peer-level conversation. As interest in AI licensing grows, many organizations are discovering that unlocking value from existing libraries requires more than just volume—it depends on structure, rights clarity, and operational readiness.
The discussion will explore how studios and broadcasters are approaching AI licensing today, where deals tend to stall, and what’s required to move from one-off transactions to more repeatable outcomes. Topics include buyer expectations, compliance considerations, and how AI licensing fits alongside traditional distribution models.
One of the quieter but more significant changes reflected at MIP London this year is how content libraries themselves are being assessed. Archives are no longer seen only as catalogues for distribution, but increasingly as data assets that can be queried, segmented, and repurposed—if the underlying rights and metadata support it.
This doesn’t mean every library is immediately suitable for AI training, nor that every deal makes sense. But it does mean content owners are being asked to think differently about how their footage is organised, understood, and valued.
Versos AI will be attending MIP London 2026 throughout the week, meeting with studios, broadcasters, and rights-holders who are navigating these questions in real time. Discussions range from early-stage evaluation—Is our content even relevant?—to more advanced operational considerations around structuring, delivery, and long-term strategy.
As AI training data takes centre stage at MIP London this year, the focus is less on bold predictions and more on understanding how this market actually functions—and how content owners can engage on their own terms.
Schedule a meeting here, or come find us at the Versos AI meeting table in IET London, Savoy Place.
