OpenAI uses provenance signals such as SynthID watermarks and Content Credentials (C2PA) to help people understand whether content was generated with OpenAI tools. Provenance signals can be helpful indicators of origin, but they are not a guarantee that content is accurate, unedited, legally owned, or presented in the correct context.
What is the C2PA standard and what does it enable?
C2PA is an open technical standard that allows publishers, companies, and others to embed metadata in media for verifying its origin and related information. C2PA isn’t just for AI-generated content. The same standard is also being adopted by camera manufacturers, news organizations, and others to certify the source and history, or provenance, of media content.
C2PA metadata can include information such as the tool or service that created a file, when it was created, and other details about its origin or history.
Learn more about C2PA.
What is SynthID and what does it enable?
SynthID is an invisible watermarking technology that embeds a signal directly into generated media. Unlike metadata, the signal is part of the content itself and may persist through some edits or transformations.
SynthID provides an additional provenance signal alongside C2PA metadata. For example, if metadata is removed from a file, an embedded watermark may still provide a signal that the content was generated with supported OpenAI tools.
What types of OpenAI-generated content include provenance signals?
OpenAI uses different provenance signals for different content types.
| Content type | Provenance signals | Notes |
|---|---|---|
| Images | C2PA metadata and SynthID watermarks | Supported images generated with ChatGPT, Codex, and the OpenAI API include both signals. |
| Audio | SynthID watermarks | Supported OpenAI-generated audio include an inaudible watermark embedded in the audio itself. |
Coverage can vary by product, model, export path, file type, and when the content was created.
Why do some content types use different provenance signals?
Different media formats preserve provenance information in different ways.
Metadata, such as C2PA Content Credentials, can carry more detailed information, but it can sometimes be removed by platforms, editing tools, or file conversions.
Watermarks, such as SynthID, are embedded into the generated media itself and may be more durable through some transformations. However, watermarks generally provide less context than metadata.
When technically feasible, using both approaches where appropriate helps provide stronger provenance signals.
How can I add a visible watermark to an image generated with OpenAI tools?
You can request a visible disclosure on images generated with ChatGPT, Codex, and the OpenAI API by prompting the model to include one. For example, when generating an image, you can say: “Include a visible OpenAI watermark in the image.”
Visible watermarks are separate from embedded provenance signals like C2PA metadata and SynthID watermarks. They may help viewers recognize that an image was generated with OpenAI tools, but they are not the same as machine-readable provenance signals.
How can I check whether content was generated with OpenAI tools?
You can visit openai.com/verify to check whether a supported image, audio file, contains provenance signals associated with content generated or exported by OpenAI tools.
After you upload a supported file, the tool will look for supported OpenAI provenance signals, such as a SynthID watermark or a trusted C2PA manifest associated with OpenAI.
If the tool finds a supported OpenAI provenance signal, it indicates that the content was generated by, or exported from, OpenAI tools.
What does a verification result mean?
If a signal is detected, the tool can help confirm that the uploaded content contains a supported provenance signal associated with OpenAI.
The tool does not confirm that the content is accurate, unedited, legally owned, or presented in the correct context. It also does not identify who created the content or why it was created.
What if no signal is found?
If no signal is detected, it means the tool did not find OpenAI issued provenance signals in the uploaded file.
The content could still have been generated or exported by OpenAI if:
The content was created before provenance signals were available.
The content came from an unsupported product, model, export path, or file type.
Metadata was stripped during upload, download, editing, conversion, or sharing.
A watermark was degraded by compression, cropping, noise, edits, format conversion, or other transformations.
An audio clip is too short or has been significantly modified.
Do you have plans to support text outputs?
Consistent with our commitments under the European Commission’s Code of Practice on Transparency of AI-generated content, our goal is to expand provenance signals to all modalities including text, so customers and developers have clear ways to meet their own transparency obligations as standards and tooling continue to mature.
Can OpenAI Verify detect content from other AI tools?
OpenAI Verify is designed to detect supported provenance signals associated with content generated or exported by OpenAI tools.
It is not designed to detect content generated by other AI services, and it may also not detect OpenAI-generated content if the relevant signal is missing, unsupported, or degraded.
Can developers verify content using an API?
Developers and organizations can use OpenAI’s verification API to check supported content for OpenAI provenance signals and integrate verification into their own applications or workflows. Visit openai.com/verify for information about API access. Organizations that need higher usage limits are able to apply for increased limits. Please visit openai.com/verify for more information.