Shutterstock-Lightricks partnership offers example of how to ethically source training data for AI video

Shutterstock’s innovative “research license” model with Lightricks marks a significant shift in how AI companies can legally and ethically access training data, potentially making high-quality datasets more accessible to startups and smaller developers.

The groundbreaking partnership: Shutterstock and AI creative technology company Lightricks have established a new licensing framework that allows AI companies to access training data through a graduated approach.

• Lightricks will train its open-source video generation model LTXV using Shutterstock’s HD and 4K video library
• The model enables companies to begin with a smaller research license for testing before upgrading to commercial licenses
• This approach directly addresses the prohibitive costs that often drive companies toward unauthorized data scraping

Financial implications and creator compensation: Shutterstock has implemented a fair revenue-sharing structure that benefits both content creators and AI companies.

• Contributors receive 20% of revenue from data licensing deals
• Content creators can opt out of AI training usage, though only 1% have chosen this option
• The model provides a legitimate alternative to unauthorized data usage, which has resulted in lawsuits against major AI companies

Technical advantages and applications: The partnership aims to solve specific challenges in AI video generation while expanding capabilities.

• Lightricks’ LTXV model focuses on improving motion consistency in longer videos
• The model has gained significant traction with thousands of downloads on GitHub and Hugging Face
• Real-time video generation for interactive ecommerce represents a key use case

Strategic significance: The collaboration represents a broader shift in how AI companies can access training data while maintaining ethical standards.

• Shutterstock has existing partnerships with major tech companies including Nvidia, Meta, and OpenAI
• The research license model could democratize access to high-quality training data for smaller organizations
• Lightricks’ decision to make LTXV open-source aligns with industry trends toward transparency

Market implications and industry impact: This new approach to AI training data access could reshape industry practices and standards.

• The model provides a viable alternative to unauthorized data scraping
• Smaller companies and startups now have a more accessible path to ethical AI development
• The success of this approach could influence other content providers to adopt similar flexible licensing models

Future outlook: As concerns about AI training data sources continue to grow, this partnership could establish a new industry standard for ethical AI development, though its long-term success will depend on adoption rates and the ability to balance accessibility with fair compensation for creators.