Google launches public SynthID detector for AI‑generated media

Google has opened its SynthID verification site to everyone, allowing users to upload images, video, or audio clips and check whether they carry the company’s AI‑generated watermark.

How SynthID works

The service accepts a wide range of formats – JPG, PNG, WEBP, AVIF, HEIC, GIF for pictures; MP4, MOV, WEBM for video; and WAV, MP3, OGG, FLAC, AAC, M4A for audio. Google reports that the tool handles about one million verification requests each day.

SynthID was first introduced in 2023 and is now embedded in Google’s own generative models such as Gemini, Veo, Lyria and the Nano Banana family, as well as in tools like Flow, ProducerAI and Vids. Partners including OpenAI, NVIDIA and Kakao have also adopted the standard, while Apple is said to be preparing support.

Why it matters for GPU / AI infrastructure

Reliable detection of AI‑created media helps build trust in digital content streams and reduces the risk of misuse. For GPU‑heavy workloads that generate synthetic assets, having a lightweight verification step can be integrated into pipelines without adding significant compute overhead, while still providing a crucial safeguard for enterprises that rely on AI‑generated visuals and audio.