Shutterstock launches a “research license” for AI companies, starting with Lightricks, which will train its video model LTXV on Shutterstock's video library
Michael Nuñez / VentureBeat :
Context & Ripple Effects
Shutterstock has been building an AI-licensing business through an expanded OpenAI training-data partnership, and its CEO said that business generated $104 million in 2023. The Lightricks arrangement extends that strategy from a major lab partnership to a repeatable research-license product.
The move also follows a competing precedent: Getty Images and Runway partnered on an AI video model for Hollywood and advertising. Video libraries are becoming a commercial input to model development rather than only a source of finished stock assets.
First-order effects
- Lightricks gains licensed access to Shutterstock's video library to train LTXV, giving its model-development effort a defined commercial data source.
- Shutterstock creates a new research-license route for AI companies, broadening its training-data licensing business beyond its earlier OpenAI deal.
Second-order effects
- Other video-model developers gain a clearer example of how stock libraries can package training access, increasing pressure on rival archives to offer comparable licensing terms.
- Licensable video data becomes more explicitly differentiated from ordinary stock downloads: AI developers may evaluate libraries for training rights and coverage, not just asset sales.
Third-order effects
- If these arrangements proliferate, stock-media companies could increasingly operate as data-rights intermediaries for generative AI, with licensing products sitting alongside traditional content distribution.
- The pattern may make provenance and rights-cleared datasets a more important competitive factor in video AI, though the corpus does not establish how broadly research licenses will be adopted.
The trend: Generative-AI developers are moving toward formal, commercial training-data partnerships with rights holders, especially for video.