AI, Meta and copyright infringement
Background
The use of Intellectual Property by AI companies’ large language models has been known for some time, but this report by the Atlantic set out the scale of scraping by Meta. Its activity was conducted via the LibGen database of pirated content, which contains more than 7 million illegally copied books and 80 million research articles. The Atlantic has produced a search tool that publishers and authors can use to discover if their content has been hosted by LibGen.
There is uncertainty about whether Meta is entitled to scrape like this under an exemption to US copyright law. However, it seems clear that it is in breach of UK copyright law as it stands.
The IPG’s responses
Like nearly everyone in publishing, the IPG views the scraping of published content by Meta and others without express permission as a completely unjustified use of Intellectual Property. We have:
- Written to Meta to protest in the strongest terms and invite a response
- Been active partners in cross-industry lobbying, including via the Creative Rights in AI Coalition (see below)
- Supported the use via PLS of collective licensing as a solution providing fair use of publishers’ content for all concerned (see below)
- Surveyed members on approaches to AI and copyright. We found the large majority are in favour of either strengthening copyright to require AI-related licensing (73%) or maintaining the copyright framework as it stands (14%)
- Used these findings to inform our detailed response to the government’s recent consultation on AI and copyright, emphasising the urgent need for robust protection of content
- Covered AI issues in depth at our Spring and Autumn Conferences
- Run award-winning training in generative AI that covers legal and ethical questions around Intellectual Property and AI, as well as the more positive use of AI tools. We also host quarterly ‘Lunch and Learn’ sessions to cover the latest AI developments, including infringement concerns.
We recognise that without major legal costs, recourse for rightsholders after Meta’s activity is frustratingly limited at the moment, but we will be every opportunity to represent members’ interests.
Copyright law and lobbying
Publishing and creative organisations have come together to protest against unauthorised use of AI. The Creative Rights in AI Coalition was set up to campaign for copyright protection and fair remuneration in the AI age. The IPG is pleased to be one of the founding members. CRAIC’s principles are here and its members are here. If you would like to protest to your MP, CRAIC has a template email requesting the safeguarding of creative rights in AI.
The Make it Fair campaign, established by a group of creative bodies, aims to educate the government and public about the need for fair remuneration for the AI-related use of Intellectual Property.
The government’s recent consultation on AI and copyright, including a proposal to introduce a new exemption to support AI-related use of content—that might enable activity like Meta’s—was a welcome opportunity to lobby for strong protection. The IPG’s submission was one of a reported 10,000+ responses that the government is now considering. We hope the government’s response will take on board the very strong feelings against the exemption that were expressed by rightsholders.
Collective licensing
Some IPG members have explored the possibility of licensing their content direct to technology firms for use by AI large language models. Our survey found that just over a third have already licensed content (9%) or had discussions about doing so (27%).
Collective licensing of Intellectual Property for use by third parties has served publishers well for the last 40 years. Publishers’ Licensing Services coordinates collective licensing in the UK and remunerates publishers for reuse of their content (the Authors’ Licensing and Collecting Society does a similar job on behalf of authors). The IPG is one of the trade associations that directs PLS.
PLS has been working with the Copyright Licensing Agency—the body managing licences to reuse content on behalf of PLS and others—to develop AI-related licences. In March it announced the addition of new permissions that allowed licence holders in the private and public sectors to use copyright-protected content to prompt generative AI tools. They come into effect on 1 May. You can sign up to receive AI-related licencing updates from CLA here.
We encourage any publisher that is not yet registered with PLS to sign up; it’s free and easy and participation is flexible.
Practical actions
Protecting content against use by Meta and others is not easy, but there are some practical steps that can be taken. For content hosted online, some IPG members have reported the effective use of robots.txt files, which tell web crawlers and AI bots which content they can and can’t visit. A plug-in for WordPress sites, Hide My WP, has also been recommended. Google has some advice for preparing these files, and web developers should be able to advise on their use.
Commercial anti-piracy solutions are available to help publishers monitor where their content is being used without permission online and issue takedown notices where necessary. IPG members providing these include Link-Busters.
For print material, publishers might consider adding a notice on the copyright page of books or journals to expressly state that content must not be used for AI training purposes.
Authors
If your authors are concerned about use of their content, publishers may like to direct them to some comprehensive advice from the Society of Authors. The Society has coordinated protests and an open letter to Lisa Nandy, Culture, Media and Sport Secretary.
Authors can search here to see if their content has been part of the LibGen database allegedly used by Meta.
Authors might like to write direct to Meta and protest that it does not have the right to use their work. The US-based Authors Guild has produced this template letter to help. They can sign the Statement on AI training that has gathered 50,000 signatures.
If you have any AI-related advice or resources that might be useful to IPG members in protecting Intellectual Property, please share them.