Just two years after holding public consultations and deciding that it was "premature" to consider amendments to the Copyright Act to address the challenges posed by the use of Artificial Intelligence (AI) enabled services, the Federal Government has launched a new consultation process.

On October 12, the Canadian government announced it was revisiting issues related to authorship of AI generated content and issues related to the use of copyright protected content in the training of AI services.

Generative Artificial Intelligence (Gen AI) refers to platforms such as Open AI's ChatGPT (a chatbot) or DALL-E (text-to-image generator). Text and images created via these tools are at times difficult to distinguish from human generated text and images. There are instances where generative AI tools and machine learning models are trained on data mined from the internet through text and data mining (TDM) activities. The tool can then recognize relationships and patterns in data. It uses this training and applies it to information submitted by end users, to produce fresh content; videos, photos and even books. Increasingly, there is significant concern over the use of publicly available content — copyrighted or not, in the training of generative AI tools.

The Ministers of Innovation, Science and Industry and Canadian Heritage want to gather stakeholder views on generative AI tools and its effects on copyright. Feedback will be collected from copyright holders on the implications of giving consent, receiving credit and compensation for the use of their works.

By way of background, as discussed in our previous blog post on Generative AI and legal considerations; Innovation, Science and Economic Development Canada (ISED) has already proposed a generative AI code of practice, thus far 14 organizations have signed on to demonstrate their commitment to responsible AI.

The government is also focusing on the implications of AI for the copyright framework.

Scope and Emphasis on Stakeholder Feedback

Recently, ISED released a consultation paper, Consultation on Copyright in the Age of Generative Artificial Intelligence which discusses the results of the government's 2021 stakeholder consultation -Consultation on a Modern Copyright Framework for Artificial Intelligence and the Internet of Things.

Three key copyright policy areas related to Artificial Intelligence were emphasized. Stakeholders were asked to provide feedback on;

  • The use of copyright-protected works in the training of AI systems - whether any clarification is needed on how the copyright framework applies to the use of copyright-protected works for training purposes;
  • Authorship and ownership rights related to AI-generated content - how the copyright framework should apply to AI-assisted and AI-generated works and;
  • Infringement and liability regarding AI - who is liable when AI-generated works infringe copyright-protected works.

From the 2021 consultation 38 submissions were received. One of the main focal points emerging from the consultation was the use of copyright-protected works in the training of AI systems. Comments from scholars and stakeholders in the technology sector argued for an exception, where the use of works in TDM would not require additional authorization from copyright holders, alternatively, stakeholders from the creative industries argued that a new exception would prevent rights holders from receiving fair compensation for the use of their works.

Legal and Regulatory Ramifications

Notably, though using copyrighted material to train generative AI models has yet to be deemed copyright infringement, there is increasing copyright litigation concerning AI platforms. Open AI is currently facing a number of lawsuits. In one copyright suit filed in California, the complainant alleges the company's AI coding assistant, Copilot, reproduced open source software without proper copyright notice.1

Also in California, Science fiction writer Paul Tremblay and novelist Mona Awad have filed suit against OpenAI alleging their works were "copied by OpenAI and ingested by the underlying OpenAI Language Model" sans permission. In the US, there is also a class action lawsuit against OpenAI alleging that the machine learning models behind ChatGPT and DALL-E, collected personal information from the internet illegally, violating state and federal privacy laws.2

Somewhat tellingly, the Federal Trade Commission (FTC) has introduced a novel disgorgement remedy, algorithmic disgorgement, which forces tech companies to terminate problematic algorithms and ill-gotten data used to train AI platforms.

The FTC's office of technology has also received feedback on the harms related to AI. Through submissions via the FTC's consumer sentinel network, the FTC identified key areas of concern for consumers.

Notably, how AI is built potentially infringing copyright by using scraped data or using biometric data such as voice prints for training, were significant points of concern for consumers.

Consumers also expressed concern about how AI works and interacts with users, bias and inaccuracies with respect to facial recognition software also troubled consumers and finally how AI is applied in the real world, aiding in producing sophisticated scams, and other malicious uses, with phishing emails becoming more difficult to detect when written using generative AI.

Given the absence of precedent, legal ambiguity and arbitrary outcomes are likely and increased regulatory scrutiny should be anticipated.

Next Steps-Feedback Accepted

As Canada continues to work toward amending itsCopyright Act, organizations in the technology and creative sectors in particular, should be familiarizing themselves with this process given the potential legal, regulatory and economic ramifications.

Given the release of dynamic generative AI tools, the current consultation is focused on improving understanding of the realities of the copyright marketplace since the government last consulted with Canadians in 2021.

Feedback generated is likely to influence copyright policy development. Stakeholders now have the opportunity to further advance the direction of Canada's copyright and AI frameworks. Stakeholder commentary will be accepted until December 4, 2023, with the aim of having findings from this consultation published in 2024.

Fasken's Intellectual Property and Technology, Media and Telecommunications groups will continue monitoring and providing updates on the results of the consultation and any updates to copyright policy.

Footnotes

1. OpenAI Legal Troubles Mount With Suit Over AI Training on Novels

2. OpenAI Legal Troubles Mount With Suit Over AI Training on Novels

The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.