OpenAI backs Europe’s AI transparency code

OpenAI announced on June 11 that it supports the European Union’s Code of Practice on Transparency of AI-generated content. The verified point is narrow: the company says it backs the framework after its publication by the European Commission and connects it directly to implementation of the EU AI Act. This is not a new model release or a flashy product feature. It is a regulatory and technical signal about a problem that now sits close to the center of AI governance: how users, platforms, and institutions can understand whether an image, video, or other piece of content was created or modified with AI.

The source focuses on provenance, meaning signals that describe where a file came from, how it was created or edited, and who signed that information. OpenAI says it has added C2PA metadata to images created or edited with DALL-E 3 since 2024. C2PA is a standard developed by a cross-industry coalition that includes media organizations, software companies, camera makers, online platforms, and AI providers. Its practical value is interoperability: content credentials can be read across multiple services instead of remaining a label inside one company’s interface.

The useful part of the announcement is that it does not treat provenance as solved. Metadata can be stripped when a file is uploaded, downloaded, resized, converted, compressed, or captured through a screenshot. OpenAI says it therefore uses a layered approach: C2PA metadata, SynthID watermarks for images generated by its products, a public verification experience, product safeguards, reporting channels, and enforcement policies. That combination does not promise perfect detection. It acknowledges that a single marker is fragile once content moves across social networks, messaging apps, news sites, and editing tools.

For Europe’s AI ecosystem, the concrete issue is turning a legal expectation into formats that can work in daily use. The Code of Practice sets an ambitious direction, but implementation will have to stay grounded in the current limits of provenance technology. OpenAI’s decision matters because a major AI provider is publicly aligning with that work while stressing that interoperability depends on many actors. Transparency for synthetic content will not be a magic switch. It is becoming shared infrastructure, built from standards, verification tools, product choices, and distributed responsibility.