This July the ASA made a submission to the Federal Government’s Joint Select Committee on Social Media and Australian Society, calling attention to the highly problematic ways in which social media companies like Meta are using authors’ and illustrators’ content to train their generative AI tools.

We were deeply concerned when Meta changed its terms of service in June of this year to allow it to use posts made on Instagram and Facebook – including creators’ copyright works – to train their AI tools. Unlike in other territories with more restrictive privacy and data laws such as the EU and UK, in Australia there is no option for users to opt out.

This poses a problem for illustrators in particular, who use Instagram to showcase their art and attract work opportunities. Now, in doing so illustrators will be improving Meta’s generative AI tools that will, in turn, substantially supplant them. In our submission we strongly objected to Meta’s terms of service which amount to a compulsory, royalty-free copyright licence that is being extracted by a market dominant player on non-negotiable terms.

Our view is that users of Instagram and Facebook must be given, as a minimum, the ability to opt out of their work being used to train Meta’s AI tools. In fact, in our submission on Safe and Responsible AI, we requested the Government insist on opt-in licensing arrangements for any AI company wishing to exploit third party copyright works for AI Training.

Our central ask to Government is that authors and illustrators – whose living is earned from their intellectual property – have the ability to control and exploit that intellectual property.

We called for the Australian Government to:

1. Require AI developers to be transparent about the content they have used for AI training. Creators should know if an AI developer seeks to use their work for AI training and must have the right to grant or decline permission. This transparency would also facilitate voluntary direct licensing and collective licensing arrangements and collaboration between the tech and creative sectors.
2. Introduce mechanisms to ensure that there is compensation to Australian creators for the unauthorised use of their works in the overseas development of Large Language Models or Multimodal Foundation Models, particularly when the models form the basis of products and services that are offered in Australia.
3. Resist lobbying from the tech sector for a new copyright infringement exception (such as a text and data mining exception) that would permit the free ingestion of copyright works for AI training.

Read our full submission to the Select Committee.