In the fast-paced world of stock media, advanced systems can automatically create images and media, opening up exciting possibilities and raising concerns about copyright, representation, and misinformation. One major player in the industry, Shutterstock, has taken a step to address these issues by introducing the TRUST framework.
Before TRUST, the stock media industry faced potential problems related to using unlicensed data for training AI systems. This raised questions about copyright infringement and fair compensation for creators whose work contributes to developing these powerful algorithms. In response to these challenges, Shutterstock has unveiled the TRUST framework, which outlines five critical ethical AI principles the company commits to follow.
To tackle the issue of unlicensed data, TRUST’s “Training” principle ensures that only correctly licensed data is used to train AI systems. This helps avoid copyright-related problems and lays the foundation for responsible AI development. Additionally, the “Royalties” principle emphasizes fair compensation for creators, with Shutterstock pledging to compensate artists through royalty funds for the use of their work in training AI models.
Representation and diversity are crucial aspects of any AI system. The “Uplift” principle of TRUST promotes diversity and inclusion in AI systems. This ensures that the generated content reflects a broad range of perspectives and avoids reinforcing biases in the training data.
Safeguarding customers and controlling AI content risks are other priorities addressed by TRUST. The “Safeguards” principle is designed to protect customers by implementing measures to prevent problematic AI content. This includes incorporating human reviews of system outputs to identify and address potential issues.
Openness plays a crucial role in establishing trust in AI-generated content. TRUST’s “Transparency” principle mandates the clear labeling of AI-generated work and supports provenance tracking. This ensures that users are informed when they encounter content created by AI, promoting transparency in using these technologies.
To demonstrate its commitment to responsible AI adoption, Shutterstock has introduced additional measures, including indemnification for copyright, trademark, and other potential risks associated with AI-generated content. This indemnification provides legal protection for customers who submit their AI-generated images for review, ensuring that they receive the same level of support and backing as traditionally licensed stock photos.
In conclusion, the TRUST framework by Shutterstock represents a significant step towards addressing the ethical challenges posed by generative AI technologies in the stock media industry. By implementing principles prioritizing responsible training, fair compensation, diversity, customer safeguards, and transparency, Shutterstock aims to lead the way in shaping the ethical development of AI in its field. As the industry evolves, the TRUST framework is a model for other companies looking to adopt ethical AI practices. It ensures that the promises of AI technology are harnessed responsibly for the benefit of creators and customers alike.
Niharika is a Technical consulting intern at Marktechpost. She is a third year undergraduate, currently pursuing her B.Tech from Indian Institute of Technology(IIT), Kharagpur. She is a highly enthusiastic individual with a keen interest in Machine learning, Data science and AI and an avid reader of the latest developments in these fields.