EU publishes its AI content labelling playbook ahead of the AI Act’s August deadline

Editor's take

The European Union has finalized its voluntary Code of Practice for AI content labelling, providing guidelines for companies to comply with upcoming transparency obligations under the AI Act. This initiative aims to equip developers and deployers of AI systems with practical steps to clearly indicate when content is AI-generated, particularly for generative AI models like OpenAI's GPT series and Meta's Llama.

This playbook's significance lies in its attempt to establish a de facto standard for AI-generated content disclosure ahead of the AI Act's enforcement. It addresses growing concerns about disinformation and manipulation, impacting consumers and democratic processes. The voluntary nature, however, raises questions about its efficacy in ensuring widespread adoption and consistent application across the diverse AI industry.

Future developments to monitor include the actual uptake by major AI providers and the effectiveness of these labels in informing end-users. The EU's willingness to adapt or mandate stricter rules, should this voluntary approach prove insufficient, will be a key indicator of the bloc's long-term strategy for AI governance and its global influence on AI transparency standards.