Creative Commons' Q1 Policy Review

We had a busy and productive first quarter when it comes to our policy work at CC. Copyright and AI remain a key focus in our work, as policymakers around the world wrestle with the difficult issues involved amid geopolitical and economic uncertainties. On the one hand, some tech companies and governments increasingly see AI development as an economic and even national security imperative. On the other hand, rightsholders, artists, and others have launched increasingly prominent protests about how AI makes use of copyrighted works. 

Our goal in intervening in these debates remains simple, if still ambitious: to ensure that the public interest and sustaining the commons are at the center of the debate.

Rather than a zero sum game where we must pick sides between competing views, we believe there are paths forward that equitably support creativity, innovation, and public interest values—and we are encouraged to see policymakers acknowledge our concerns within the policy and legislative frameworks being developed. 

Copyright, “Opt-Outs”, and AI Training

Over the last several months, CC has engaged with partners in countries that are considering whether and how to craft a legal requirement to follow copyright “opt-outs” and other preference signals for AI training. In Europe, existing copyright law already has exceptions that allow text and data mining (TDM), including AI training, and the European Commission is now developing a Code of Practice for general purpose AI developers in light of the AI Act, which will seek to clarify how developers should respect “opt-outs.”

Meanwhile, the UK government has proposed creating a new copyright exception, subject to an ability for rightsholders to opt-out, similar to the regime established in the EU by the 2019 Copyright Directive.

CC’s View

When it comes to the role of copyright within AI, we continue to believe that copyright should be limited to facilitate TDM. Studying and analyzing materials to derive facts, ideas, and other uncopyrightable elements or make other non-infringing uses should be permitted, even where that analysis requires making a copy as an intermediate step. At the same time, a copyright exception can be complemented by ways for rightsholders to express their preferences about use of their works in TDM, including in AI training. You can learn more about our position in our submission to the UK in our LinkedIn post.

CC’s Q1 Engagements & Policy Priorities 

In both our written submissions and in our meetings with policymakers in Brussels, we have stressed the need for balanced, workable solutions. Specifically, we have advocated for policymakers to support preference signals that would offer creators a simple mechanism to make flexible and granular choices.

We believe that a binary opt-in or opt-out system of contributing content to AI models is not nuanced enough to represent the spectrum of choice a creator may wish to exercise.

What is more, even where following opt-outs are legally required, policymakers should ensure that researchers and other actors with a legitimate interest can continue to engage in TDM and that opt-outs do not override TDM and other exceptions and limitations to copyright. 

In addition, we’ve continued to focus our advocacy on supporting open source and open science in AI. For instance, we worked with a coalition including EleutherAI and the Open Source Initiative on a letter cautioning against language in the EU’s Code of Practice that would require control over downstream users in ways that are not feasible for open source developers.

Looking Ahead

Looking ahead, this debate is unlikely to be concluded anytime soon. The EU Code of Practice will be finalized soon, but it will give rise to further stakeholder dialogue and standards development as the AI Act becomes fully applicable. The UK government received over 11,000 responses to its consultation, and is also keen on finding workable ways to implement preference signals. And around the world, policymakers are watching these and similar proceedings closely as they determine how to make sure copyright is fit for purpose.  

Clearly there is lots more in store for us this year, especially in the EU as we continue to engage and seek to fine-tune the copyright and transparency aspects of the EU’s technical work on the AI Act. We will continue to provide input to support the UK in its efforts to update and modernize its copyright framework. 

Finally, looking internationally, we aim to hold conversations with our community and with policymakers around preference signals as a means of ensuring the public interest and sustaining the commons remains at the heart of policy discussions the world over.