OpenAI joins EU transparency code, commits to AI-content watermarks

OpenAI signed the European Union's Code of Practice on AI content transparency this week, joining a voluntary framework designed to help users identify AI-generated content. The company commits to advancing provenance standards and deploying tools that surface when content was made by an AI system.

The code is part of the EU's broader push to build what regulators call a "trustworthy AI ecosystem." OpenAI's participation is voluntary — the code is not a binding regulation — but it signals alignment with the EU's approach to content labeling ahead of the AI Act's full enforcement timeline.

What stands out

• Provenance tooling. OpenAI commits to technical standards that embed metadata in AI-generated images, video, and audio, with the goal of machine-readable watermarks that persist through downloads and edits.
• User-facing labels. The company will surface AI-generation notices in user interfaces where its models produce content, though the code does not specify which products or how prominently.
• Voluntary compliance. The code is not legally binding under the AI Act, but signatories agree to annual reporting on their implementation. OpenAI joins other major AI vendors in the framework.
• No enforcement mechanism. The code relies on self-reporting. There are no fines or penalties for non-compliance, unlike the AI Act's tiered liability structure for high-risk systems.
• Timeline unclear. OpenAI did not publish a deployment schedule for the provenance tools or say which models will be covered first.