OpenAI outlined a multi-layered approach to content provenance that combines Content Credentials, C2PA conformance, and Google SynthID watermarking. The company published the plan on May 19, 2026 and provided updates on July 31 and October 5, 2026.
C2PA conformance and Content Credentials
OpenAI said it has been applying Content Credentials to images since 2024, first with DALL·E 3 and later to ImageGen and Sora. The company joined the Steering Committee of the Coalition for Content Provenance and Authenticity and has taken the step of becoming a C2PA Conforming Generator Product. According to OpenAI, C2PA’s metadata and cryptographic signatures help provenance travel with content and give platforms a trusted way to read, preserve, and pass along that information.
SynthID watermarking and resilience
OpenAI described metadata as a key foundation that can nonetheless be stripped or lost through common transformations. To improve signal durability, the company is adding Google DeepMind’s SynthID watermarking for images generated via ChatGPT, Codex, or the OpenAI API. OpenAI noted that watermarking can survive some transformations—such as screenshots—while metadata can carry more contextual detail. The two layers are presented as complementary rather than interchangeable.
Public verification tool and scope
OpenAI previewed a public verification tool that checks uploaded media for Content Credentials and SynthID signals to determine whether it was generated on ChatGPT, the OpenAI API, or Codex. The company said the tool will not draw definitive conclusions when signals are absent, since provenance can be removed. At launch, verification is limited to content generated by OpenAI, with plans to support cross-industry efforts and more content types over time.
In the July 31, 2026 update, OpenAI said supported audio generated with OpenAI tools will include SynthID watermarking and that the verification tool will gain audio verification and API access for developers. The October 5, 2026 update said the work will expand to text in the EU to comply with the EU AI Act.
OpenAI attributed the initiative to building interoperable provenance across platforms and described the combined use of standards, watermarking, and verification tooling as the intended path to improve how people detect and interpret AI-generated content.
Original source: OpenAI News