Google Launches AI Watermark Detector for Everyone

Google has opened its AI watermark-detection platform to the public. This comes as worries rise about fake images, videos, and audio ahead of Nigeria’s 2027 elections.

The company shared this news in a statement sent to PREMIUM TIMES. Now, the SynthID Detector is available worldwide in English.

This platform lets users upload media files. They can check if these files have invisible watermarks from supported AI-generation systems.

Currently, it can spot SynthID watermarks in content made with tools from Google and other companies like OpenAI, NVIDIA, and Kakao. Apple is expected to join this effort soon.

Users can access this service using a Google, Apple, or OpenAI account, according to the statement.

Google mentioned that the platform targets everyday users. This includes journalists, fact-checkers, and the public who need help verifying suspicious online content.

Election Relevance

The public launch comes as political activity heats up before Nigeria’s 2027 elections. Concerns are rising about the potential use of AI to create fake political statements, impersonate candidates, and sway public opinion.

AI tools can create realistic photos, cloned voices, and videos that show people doing or saying things that never happened. Such content can spread quickly on social media and messaging apps, sometimes reaching large audiences before journalists or officials can trace its source.

The Independent National Electoral Commission has flagged the use of AI-generated fake news and social media chaos as threats to Nigeria’s election credibility and security.

Google believes this detector can give Nigerian journalists and voters another way to verify questionable campaign videos, fake speeches, false endorsements, or altered audio recordings.

How It Works

SynthID is not your regular visible label or logo on a photo or video.

Instead, it adds a digital signal directly into AI-generated content when it is created. This watermark is made to be undetectable to the human eye but can be found by Google’s detection technology.

For images and videos, Google says the watermark remains detectable even after common changes like cropping, filtering, or compression. For audio, the signal is embedded so that listeners cannot hear it and can survive changes like added noise or speed adjustments.

Once a file is uploaded, the detector checks for a recognized watermark. It will then indicate if supported AI technology created or modified the content.

Google first launched SynthID in 2023. They tested an early version of the detection tool with journalists and media experts. Now, it is available for the general public.

Not a Universal AI Detector

Even though this tool is helpful, it cannot identify the source of every piece of AI-generated content online.

SynthID mainly checks for watermarks from Google and partner tech companies. Content from unsupported AI models or without a compatible watermark might not be recognized.

Therefore, if the detector finds no SynthID watermark, it should not be taken as proof that the file is real or made entirely by humans.

The detector also does not confirm if events shown in real footage are accurately described. A genuine photo or video can still be shared with the wrong date, location, or caption.

Because of this, journalists and other users will need to combine the tool with other verification methods. This includes checking sources, looking at metadata, doing reverse-image searches, comparing landmarks, and contacting relevant people or organizations.

Scale of AI-Generated Material

Google claims it has added SynthID watermarks to over 180 billion AI-generated images and videos since launching the technology.

It has also watermarked around 240,000 years of generated audio.

The company’s other verification tools are already part of Google Search, Gemini, and Chrome. Google says these services handle over one million media-verification requests daily.

SynthID also allows watermarking for text produced through certain Gemini products. However, Google’s public detector is mainly focused on verifying images, videos, and audio files.

Google said this initiative aims to give users better context about the online content they see. For Nigeria, the platform provides another way to check information at a time when election-related fake news is growing more complex. It does not replace traditional fact-checking but can help users spot some synthetic media before sharing or acting on it.