Executive Summary
OpenAI has officially announced its support for the European Commission’s new Code of Practice on Transparency of AI-Generated Content. This commitment is framed as a key step in implementing the EU AI Act and fostering a more trustworthy digital ecosystem. The company highlighted its multi-layered approach to content provenance, including C2PA metadata and watermarking, as its contribution to this collaborative, ecosystem-wide effort to provide context for AI-generated media.
Key Takeaways
* Official Endorsement: OpenAI formally supports the EU's new voluntary "Code of Practice on Transparency of AI-Generated Content."
* Multi-Layered Provenance: The company's approach to transparency combines several techniques, including C2PA Content Credentials for metadata, SynthID watermarks for resilience, and a public verification tool.
* Existing Implementation: OpenAI has been adding C2PA metadata to images from DALL-E 3 since 2024 and now applies both metadata and watermarks to images generated via ChatGPT and its API.
* Ecosystem Collaboration: The announcement stresses the need for industry-wide cooperation, referencing OpenAI's participation in the C2PA Steering Committee alongside device manufacturers, media organizations, and platforms.
* Acknowledged Challenges: OpenAI recognizes that provenance is a nascent field and that signals like metadata and watermarks can be lost or removed as content moves across the internet.
Strategic Importance
This announcement positions OpenAI as a cooperative partner with EU regulators, proactively aligning with transparency principles ahead of the full implementation of the EU AI Act. It reinforces the company's public image as a proponent of responsible AI development and governance.