gen‑ai.news
← Back
Multimodal

Google’s new SynthID website can identify AI-generated media

Google has opened up its SynthID watermarking technology to the general public through a dedicated website, where users can upload media files to determine whether they were produced by an AI system. The tool supports images, video, and audio, making it one of the broader consumer-facing detection utilities available from a major AI developer.

SynthID was originally developed by Google DeepMind and has been embedded in several of Google's own generative AI products, including Imagen and Veo. The system works by embedding imperceptible watermarks directly into the pixels of an image, the frames of a video, or the waveform of an audio file - signals that persist even after common edits like cropping, compression, or brightness adjustments. Until now, SynthID was primarily accessible through Google's APIs and select partners rather than as a standalone consumer tool.

The move to a public website lowers the barrier for journalists, researchers, and general users who want a straightforward way to scrutinize media they encounter online. As AI-generated content becomes harder to distinguish visually from real footage, detection tools are increasingly important for maintaining a basic level of media literacy. That said, SynthID can only reliably detect watermarks it placed itself - content generated by other AI systems, or AI content that has been heavily manipulated, may not be flagged.

This launch fits into a broader industry push toward provenance and transparency standards, such as those being developed by the Coalition for Content Provenance and Authenticity (C2PA). Google has been a participant in those efforts, and SynthID complements that work by offering a watermark-based layer of verification. Making the tool publicly accessible suggests Google is positioning SynthID as infrastructure for trust rather than just a backend feature, though its real-world effectiveness will depend heavily on how widely its watermarking gets adopted across AI content pipelines.

Enjoy this story? Get the next one in your inbox.

Twice a week: the most important stories in generative image and video AI, distilled into a 2-minute read.

Free. Unsubscribe any time. No spam, ever.

Your next read

Reka AI's omni-model Rho-1 handles text, images, video, and robot control in a single model
Multimodal

Reka AI's omni-model Rho-1 handles text, images, video, and robot control in a single model

Reka AI has released Rho-1, a 19-billion-parameter model that handles text, images, video, and robot control actions within a single neural network. Rather than routing different tasks to specialized subsystems, it processes all modalities as tokens in one shared context window. The model was trained on 320 H100 GPUs over roughly three months, using considerably less compute than many leading models today.

Uniting LED Volumes and Live Visual Chaos: Cinematographer Bradford Lipson on 'Rolling Loud'
Multimodal

Uniting LED Volumes and Live Visual Chaos: Cinematographer Bradford Lipson on 'Rolling Loud'

Cinematographer Bradford Lipson explains how he and director Jeremy Garelick used LED volume technology to blend real footage from the Rolling Loud music festival with controlled stage work, creating the illusion that actors were surrounded by tens of thousands of live concertgoers. The production shot on location at the festival for just three days before rebuilding that world on the Lux Stage at Trilith Studios. Lipson walks through the technical and artistic decisions behind matching concert

AMD Acquires World Labs for $8.2 Billion, Bringing Fei-Fei Li Onboard
Multimodal

AMD Acquires World Labs for $8.2 Billion, Bringing Fei-Fei Li Onboard

AMD has agreed to acquire World Labs, the spatial intelligence startup founded by AI pioneer Fei-Fei Li, for $8.2 billion. Li will join AMD as Executive Vice President and Chief Scientist, reporting to CEO Lisa Su. The deal gives AMD an in-house AI research unit built around world models that generate and simulate 3D environments.