Concerns regarding copyright have prompted photography agencies to remove images generated by intelligence models from their databases. The concerns arise because AI models consume amounts of copyrighted human generated artwork to train themselves utilizing this database to produce images based on various text prompts. As a result the question of who owns the copyright, for an image created by an AI has sparked debate.
Popular stock photography agencies like Getty Images and Shutterstock recently announced their decision to eliminate images generated by AI models such as DALL E, Stable Diffusion and Midjourney.
According to a spokesperson from Getty Images “there are questions and genuine concerns surrounding the copyright of outputs produced by these models.”
Despite assertions from AI model creators, including OpenAI, that the copyright for generated output belongs to the user who inputs the prompt into the AI system smaller photo agencies now face pressure to follow suit with companies. Dittmar Frohmann from Photocase, a stock agency stated that his company is likely to adopt a stance.
Ryan Abbott from the University of Surrey in the UK states that “simply removing AI generated images entirely from a platform does not provide a solution, for any of these challenges.”He mentions that while it is theoretically possible to copyright images produced by an AI, in the UK the same possibility has not been tested in practice. The US Copyright Office also follows a policy of not granting copyright to AI generated art.
According to Neil Brown, a lawyer at Decoded Legal, the Copyright, Designs and Patents Act of 1988 in the UK protects all works that’re computer generated ” which would likely include AI generated works. However there are complexities involved. Not could the resulting image be subject to copyright protection. Even the input instructions could potentially be protected if they are detailed enough.
Online platforms have already emerged where people can buy and sell text prompts. If someone can provide evidence that their text prompt is being used by someone without permission they may have grounds, for a copyright claim. Brown points out that the terms and conditions set by AI creators can further complicate matters.
In my opinion it is possible for your created work to be protected by copyright; however this matter is still unsettled. Could potentially lead to legal disputes.
According to Frohmann Photocase is likely to ask for permission, from the photographers who submit images to their agency before allowing AI companies to use their archive for training their AI models. This precaution is taken because some artists may have concerns about their work being utilized to develop an AI that could potentially replace them. Larger companies like Getty are already incorporating references to AI training in their release forms.
Despite his stance Frohmann views AI as a tool to editing software. Believes that the copyright should rest with the person who prompts the AIs actions. Moreover he currently does not perceive any threat posed by AI art towards peoples livelihoods.
In Frohmann’s words “If you consider everything together its mostly uninspiring. The images might appear pleasing, at glance but lack originality and a sense of soul.”
