Shutterstock Builds on Data Licensing Strength with New AI Services for Model Training and Evaluation

Core Insights - Shutterstock has launched a suite of AI services aimed at model builders, providing specialized training datasets and evaluation tools to enhance the development of next-generation AI systems [1][3] - The company is transitioning from a data provider to an active partner in model construction, training, and assessment, leveraging its extensive content corpus and metadata licensing business [1][4] Group 1: AI Services Overview - The new AI services will deliver fast, high-impact, custom content and insights tailored to the specific needs of AI models, aligning with human preferences and creative intent [1][3] - Shutterstock's AI services include custom datasets, human-centered models, and enhanced AI capabilities, addressing the growing demand for high-quality, human-calibrated signals throughout the training lifecycle [2][3] Group 2: Strategic Partnerships and Collaborations - Shutterstock has established itself as a valuable partner for industry leaders such as NVIDIA, Meta, OpenAI, and Runway, contributing to the development of generative AI tools and standards across various media formats [4][5] - The company’s collaborations are focused on shaping the next generation of AI technologies, utilizing its rich content and extensive metadata [4][6] Group 3: Comprehensive Support for Model Training - Shutterstock offers end-to-end support for model training, including custom content production, creative-expert preference data, curation and annotation services, and evaluation and performance insights [4][8] - The company operates 10 production hubs with over 2 million creators across more than 150 countries, enabling the delivery of structured datasets that include production notes and metadata [8] Group 4: Unique Value Proposition - Shutterstock's unique position at the intersection of data, creativity, and technology allows it to provide model builders with essential resources to create safer and more capable AI systems [6][8] - The company’s advanced machine learning tools and human reviewers facilitate trait-specific scoring and annotations, enhancing the quality of training examples for AI models [8]