OpenAI expands verification tools as AI slop blurs digital trust
Synthetic media verification is becoming a larger priority as OpenAI expands provenance systems.
OpenAI has announced new measures to strengthen the provenance and verification of AI-generated content as synthetic media becomes more widespread across digital platforms.
The company said it is expanding support for Content Credentials and compliance with the Coalition for Content Provenance and Authenticity (C2PA) standard. The standard uses metadata and cryptographic signatures to help ensure that information about a piece of media travels securely with the content, including details on where it came from and how it may have been created or edited.
OpenAI also plans to integrate Google DeepMind’s SynthID watermarking into images generated through ChatGPT, Codex and the OpenAI API. The company said SynthID will add an invisible watermarking layer that complements C2PA metadata, particularly when metadata is removed, lost, or altered during file conversions, resizing, screenshots, or other transformations.
The company said it is adopting a multi-layered provenance approach that combines metadata, watermarking and public verification tools rather than relying on a single detection method. According to OpenAI, C2PA can provide richer contextual information, while SynthID can help preserve a signal when metadata does not survive.
The move also connects to wider concerns about AI slop, as synthetic media and low-quality AI-generated content become harder to distinguish from authentic images. Provenance tools cannot solve the problem alone, but they can provide clearer signals about how digital media was created or modified.
OpenAI also previewed a public verification tool that will allow users to check whether ChatGPT, Codex or the OpenAI API generated an uploaded image. The tool will look for provenance signals, including Content Credentials and SynthID watermarks. Still, OpenAI said it will not make a definitive judgement when no signal is detected, because provenance signals can sometimes be removed.
At launch, the verification tool is limited to OpenAI-generated content. The company said it aims to support wider cross-platform verification efforts in the coming months and eventually expand support to more types of online content.
Why does it matter?
AI-generated content is becoming harder to distinguish from authentic media, fuelling concerns around AI slop, deepfakes and manipulated information. Provenance systems such as Content Credentials, watermarking and verification tools can help people understand where media came from and whether it was generated or modified by AI. However, OpenAI’s approach also shows the limits of technical detection: metadata can be stripped, watermarks may not survive every transformation, and no single method can provide complete certainty.
Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!