In an era where distinguishing between human and AI-created content has become increasingly challenging, Google has introduced a powerful new verification tool at its I/O 2025 developer conference.
The SynthID Detector portal enables users to quickly identify AI-generated content by scanning for invisible watermarks embedded by Google's AI systems. When users upload an image, audio track, video, or text created with Google's AI tools like Gemini, Imagen, Lyria, or Veo, the portal not only determines if the content contains a SynthID watermark but also highlights specific portions most likely to be watermarked.
This technology addresses a critical need as AI-generated media floods the internet. According to recent estimates, deepfake videos increased by 550% between 2019 and 2024, while a significant percentage of highly-viewed social media posts now contain AI-created content. SynthID's watermarking system remains detectable even when content undergoes transformations like cropping, resizing, or other modifications, making it particularly effective against attempts to disguise AI origins.
Google has already expanded the reach of SynthID beyond its own ecosystem through strategic partnerships. The company has open-sourced its text watermarking technology and partnered with NVIDIA to watermark videos generated by NVIDIA's Cosmos model. Additionally, a collaboration with GetReal Security will enable external verification of SynthID watermarks, helping build a more trusted digital ecosystem.
While SynthID Detector represents significant progress in content verification, it does have limitations. The tool primarily works with Google's AI models and those using its watermarking technology, leaving content from other platforms undetectable. Critics also note that watermarking doesn't prevent content misuse but only enables after-the-fact detection.
The SynthID Detector is currently available to a select group of early testers, with journalists, media professionals, researchers, and educators able to join a waitlist for access. A broader rollout is expected in the coming weeks as Google continues to refine this important tool in the fight against AI-generated misinformation.