Photo by Negative Space from pexels.com
Canadian creators and publishers are pressing the government to address the unauthorized use of their content in training generative artificial intelligence systems. They argue that this practice infringes on copyright and undermines their ability to control how their work is used. On the other hand, AI companies contend that such use is essential for advancing AI technology in Canada and should not be restricted.
These opposing viewpoints are being presented in submissions to a consultation on copyright and AI by the federal government. The consultation aims to update Canada’s copyright laws to accommodate emerging technologies like OpenAI’s ChatGPT, which can generate text, images, videos, and code based on extensive datasets it has learned from.
Access Copyright, representing writers, artists, and publishers, describes the current use of their works by AI systems as occurring in a "black box." They highlight that while rightsholders are aware of this activity, they lack the means to identify which works are being used or to prevent unauthorized use.
Music Canada, representing major record labels, pointed out a specific incident where an AI-generated song mimicked popular artists without permission, illustrating the extent to which proprietary data is being ingested without authorization.
The Writers’ Guild of Canada has called for basic disclosure and reporting obligations, emphasizing the disparity in information between developers, who have detailed knowledge of how content is utilized, and creators, who are often left uninformed.
While some organizations have reached licensing agreements with AI companies, the Canadian Authors Association argues that many rightsholders face significant challenges in licensing their content due to lack of transparency about its use.
Legal actions in the United States highlight these transparency issues, with cases where AI companies argue that rightsholders cannot prove infringement without specific details about the training data used.
The Canadian Media Producers Association stresses the urgency of addressing these issues, urging the government to implement transparency requirements to protect rightsholders.
AI companies, however, argue that the level of transparency demanded by rightsholders is impractical. Microsoft, for example, asserts that recording and disclosing vast amounts of training data would hinder AI development.
Google echoes this sentiment, stating that AI training is already exempt from copyright law and that additional regulations could impede innovation by exposing sensitive competitive information.
Cohere, a Canadian AI firm, likens using content for AI training to reading books for knowledge acquisition, asserting that it does not constitute copyright infringement and should be clarified as such in the law.
The Council of Canadian Innovators warns that stringent disclosure rules could disproportionately harm smaller tech companies compared to larger competitors, potentially stifling their growth and innovation.
As Canada navigates these complex issues at the intersection of AI and copyright law, the outcome of the consultation will likely shape the future landscape for both creators seeking protection and AI companies aiming to innovate.
