The European Commission is drafting a new Code of Practice (CoP) for General-Purpose AI. This initiative aims to align AI practices with the upcoming EU AI Act. The CoP will focus on transparency, risk assessment, copyright compliance, and systemic risk taxonomy. However, the draft has drawn criticism, particularly regarding its potential impact on copyright protections.
Copyright Groups Sound the Alarm
Fifteen European rightsholder groups, including News Media Europe and the Federation of European Publishers, have voiced strong concerns. In a letter to Henna Virkkunen, the European Commissioner for tech sovereignty, they warned that the CoP could weaken existing EU copyright laws.
The controversy stems from a proposed requirement for AI providers. They are asked to make “reasonable and proportionate efforts” to ensure legal access to copyrighted materials. Critics argue this change undermines the EU’s current copyright rules, which demand clear and lawful access for AI training.
News Media Europe criticized the draft’s strict interpretation of text and data mining rules. They claim it could limit rightsholders’ ability to verify the use of their content. Furthermore, they see this approach as a potential “backdoor for legislation.”
The draft also introduces a transparency template for AI providers. However, rightsholder groups argue that it falls short of achieving the AI Act’s goals and may even hinder them.
Balancing Flexibility and Enforcement
Officials drafting the CoP aim to make it adaptable for various industries, such as music and publishing. However, critics worry this flexibility could reduce enforcement and harm copyright protections.
The next draft of the CoP will be released in mid-February. Meanwhile, the AI Act is set to take full effect in August. Some provisions, like bans on facial recognition systems, will begin as early as February 2.
Global Echoes of Copyright Debates
The EU’s challenges with AI and copyright reflect similar global disputes. In the United States, lawsuits have already been filed over alleged copyright violations by AI companies. For instance, in 2023, Sarah Silverman sued Meta and OpenAI. She claimed her memoir was used illegally to train their AI systems.
These debates highlight the complexity of regulating AI while respecting intellectual property rights. As the EU finalizes its CoP and AI Act, its decisions may set a global precedent for balancing innovation and copyright protection.
