OpenAI announced in May 2026 two new measures to help users identify images generated by its AI models: support for an open metadata standard called C2PA and a partnership with Google to embed an invisible watermark called SynthID into generated images.
The C2PA standard, developed by the Coalition for Content Provenance and Authenticity — a non-profit founded in 2021 — embeds a readable signal directly into a file’s metadata indicating that an image was AI-generated. Because the signal is easily accessible, it can also be manipulated, making it most reliable among trusted users. SynthID, developed by Google, takes a different approach: its invisible watermark is designed to persist even when someone attempts to remove it through screenshots, resizing, or other digital manipulation.
OpenAI described the two systems as complementary. “Watermarking can be more durable through transformations like screenshots, while metadata can provide more information than a watermark alone,” the company said in its announcement. “Together, they make provenance more resilient than either layer would be on its own.”
OpenAI is also previewing a public verification tool that will check images for both signals, allowing users to test whether a given image was produced by an OpenAI product. The company says it hopes to expand the tool to cover images from other AI platforms over time.
The protections currently apply only to images generated through OpenAI’s own products and will not affect imagery produced by other AI tools. However, the move may help ensure OpenAI is not contributing to the broader challenge of AI-generated images circulating without clear provenance — a problem that has grown as AI image generators have become more widely available and harder to distinguish from authentic photography.
Source: TechCrunch