Stability AI denied accusations of copyright infringement by Getty Pictures, calling them “vague” and “general”. This is the latest conflict between a tech company and a creative firm over intellectual property.
Stability, in a defense filed before the High Court, denied that Getty images available online are “highly desired” for training their artificial intelligence models, as they have watermarks and poor resolution. The company said that the pictures created by the technology “did not derive any output” from the entire or substantial part of the copyrighted works.
Stability’s lawyers said that “billions” of images of similar or higher-quality, with alt text captions of equivalent or higher-quality, were widely available in other sources, and the models had been trained outside the UK.
Stability also said that Getty’s proof that images were generated by AI was weak, contrived, and that technology was told to produce something similar.
Stability AI, a start-up based in London, was founded by Emad Mostaque in 2019. It focuses on generative AI. It can create text, images and music from human prompts. Getty Images has sued it in the United States and at the High Court over copyright. They claim that the machine illegally scraped images from the website of Getty Images to train it.
Stability acknowledged that it had datasets that included links to pictures from Getty Images, but did not specify the extent or whether these images were copyrighted.
Tensions between the technology industry and creative industries regarding copyright are rapidly increasing. To train AI models, large pools of high-quality data are needed. However, data owners want to be compensated for their use. It is difficult to discover what data is being used.
Several major court cases were initiated. The New York Times is suing Microsoft and OpenAI. OpenAI faces a lawsuit from the Authors Guild that includes John Grisham, Jodi Picoult and others. Universal Music has sued Anthropic for the use of song lyrics.
The UK Department for Science, Innovation and Technology was tasked to create a code on copyright and AI, which aims “to increase the availability of licences for data-mining.” It will help overcome the barriers AI users and firms currently face, and ensure that there are rights holders’ protections.”
Stability provides an opt-out option for companies who do not wish to have their data used. It said that many of these companies had done so since last May.
Getty Images has declined to comment.