Artistic licence: can copyright rein in generative AI?
When artist and illustrator Kelly McKernan discovered that her artwork had been used to train the artificial intelligence (AI) art tool, Stable Diffusion, she initially found it “surreal and exciting”.
But as she later revealed in a tweet, this elation swiftly turned to “nauseating” anger—prompting a lawsuit that could have crucial implications for both AI developers and copyright owners.
After realising that her “art had been trained on without her consent or compensation”, McKernan, and fellow artists Sarah Andersen and Karla Ortiz, brought a class action complaint against the tool’s creator, Stability AI, earlier this month.
The case marks the start of a flurry of litigation against the developers of generative AI— systems that can learn concepts from large bodies of existing knowledge and which then use what they learn to help people create new works.
In London, Getty Images sued Stability AI for using images for training purposes without its permission, while Microsoft, Microsoft’s GitHub, OpenAI and MidJourney have been accused of infringement by allegedly using copyrighted open source code to train their machine learning (ML) systems.
International repercussions
But according to lawyers and academics, it is the artists’ case that has the power to reverberate across the legal sphere, and to truly divide opinion.
For Mark Nichols, senior associate at Potter Clarkson, the US case covers more territory than the Getty lawsuit, as it explicitly looks at the IP implications regarding content that has been created by an AI—even if it has used the work of others to do so.
“It asks whether the output from an AI infringes as a derivative work, whereas the Getty case just concerns the content that was an input,” he explains.
