OpenAI Launches C2PA, SynthID, Image Verification
OpenAI announced three linked updates to its AI content provenance system: formal C2PA conformance, integration of Google DeepMind's SynthID watermarking for images, and a preview of a public tool that lets users verify whether an image was generated by ChatGPT, Codex, or the OpenAI API. According to the company's announcement, the goal is to make AI-generated media easier to identify and trace across platforms even when one layer of provenance signal is lost.
How C2PA Conformance Helps Provenance Travel Across Platforms
OpenAI said it has become a C2PA Conforming Generator Product, a designation from the Coalition for Content Provenance and Authenticity. The C2PA standard uses metadata and cryptographic signatures that travel with a file, providing context about how media was created or edited.
The company stated that conformance gives other platforms a trusted way to read, preserve, and pass along provenance information attached to OpenAI-generated content. Without that interoperability, Content Credentials can become siloed on the platform where an image was first created.
OpenAI noted it has supported provenance standards since 2024, when it began adding Content Credentials to DALL·E 3 images and later extended them to ImageGen and Sora. The company also said it joined the C2PA Steering Committee.
Why OpenAI Added Google SynthID as a Second Layer
According to OpenAI, C2PA metadata is not foolproof on its own. It can be stripped during uploads, broken by file format changes, or lost through screenshots and resizing. To address that vulnerability, the company said it is now embedding Google DeepMind's SynthID watermark into images generated through ChatGPT, Codex, and the OpenAI API.
OpenAI described SynthID as an invisible watermarking layer that complements metadata-based approaches. The company said the two systems reinforce each other: C2PA carries richer context about creation history, while SynthID preserves a detectable signal when that metadata does not survive normal sharing workflows.
The company noted it has previously used visible watermarks in Sora and an audio watermark in Voice Engine, and has continued testing accuracy and reliability over time.
What the Public Verification Tool Does for Users
OpenAI is also previewing a public verification tool at openai.com/verify. The company said users can upload an image and the tool checks for both Content Credentials and SynthID signals to determine whether the image came from OpenAI's systems.
OpenAI cautioned that detection is not definitive in every case. If the tool does not find metadata or a watermark, it will not conclude the image was not made with OpenAI tools because provenance signals can sometimes be removed or degraded.
At launch, verification covers only content generated by OpenAI. The company said it aims to support cross-industry verification efforts and additional content types in the coming months.
What This Means for Identifying AI-Generated Media
OpenAI's layered approach open standards, durable watermarking, and public tooling gives journalists, platforms, and everyday users multiple ways to check whether an image is AI-generated. The combination means that even if metadata is stripped during social media uploads, the SynthID watermark can still be detected through the verification tool.
The current rollout is limited to OpenAI-generated content, so it does not yet serve as a universal detector for all AI imagery. But according to the company, broader cross-platform support is planned. For users who want to check images now, the verification tool is accessible and requires only an image upload.