Google launches SynthID Detector, which lets users identify AI-generated image, video, and audio from Google, Nvidia, OpenAI, and Kakao across 20 file formats
Google on Tuesday launched a new site that lets anyone verify whether a piece of media, be it an image, a video or an audio clip, is generated using AI.
Context & Ripple Effects
SynthID began as an invisible watermark for Google-generated images in 2023, then expanded to text-generation watermarking and detection in 2024. OpenAI’s 2026 support for the scheme moved it beyond a single vendor’s tooling and toward shared provenance infrastructure.
Google’s watermarking work has also broadened into scientific content through SynthID Bio. But prior testing found the watermark difficult to remove without solving the broader problem that unlabeled AI media can still be created.
First-order effects
- Anyone can submit supported image, video, and audio files to a public verification endpoint rather than relying on developer-facing watermark tools.
- Google says the service includes checks for participating partners including OpenAI, NVIDIA, and Kakao, making their supported media easier to verify through one interface.
Second-order effects
- OpenAI’s earlier adoption of SynthID gains a consumer-facing use case: provenance markings can be checked by audiences, not only embedded by model providers.
- AI vendors that do not support interoperable provenance signals face a clearer comparison with partners whose labeled output can be verified publicly.
Third-order effects
- The move strengthens a synthetic-media trust stack built around embedded provenance plus accessible verification, rather than watermarking alone.
- Because watermarking does not identify media made without a label, public verification is likely to become one layer of authenticity assessment rather than a universal AI-content detector.
The trend: AI provenance is moving from vendor-specific watermarking toward interoperable, public-facing verification across media formats.