Google has made its artificial intelligence (AI) watermark detection platform, SynthID Detector, available to the public globally, giving journalists, fact-checkers and other users a tool to examine images, videos and audio for signs of AI-generated content.
The move comes amid growing concerns about deepfakes and online misinformation ahead of Nigeria’s 2027 general elections.
Google DeepMind announced the wider availability of the tool on October 7, 2026, extending access beyond the journalists and media professionals who previously tested it.
According to the company, SynthID Detector supports AI-generated media from Google and participating technology companies, including OpenAI, NVIDIA and Kakao. Apple is expected to join later. Users can access the platform with a Google, Apple or OpenAI account.
How SynthID Detector Works
SynthID embeds digital watermarks into AI-generated content during creation. Unlike visible labels or logos, the watermarks are designed to remain imperceptible to humans while allowing compatible detection systems to identify them.
Google said the watermarks are designed to remain detectable after common alterations. In images and videos, they can survive cropping, filters, compression and changes to video frame rates.
Audio watermarks are inaudible to listeners and are designed to withstand modifications such as added noise, speed adjustments and MP3 compression.
When users upload a supported file, the detector scans it for a recognised SynthID watermark and reports whether it finds evidence associated with supported AI technologies.
Google introduced SynthID in 2023 to help identify AI-generated media. The company said it has since watermarked more than 180 billion images and videos, alongside audio equivalent to about 240,000 years.
Concerns Over AI Misinformation Ahead of 2027 Elections
The platform’s public release comes as concerns grow over the potential use of AI-generated content to impersonate political candidates, fabricate endorsements and spread misleading statements during Nigeria’s 2027 election cycle.
Generative AI tools can produce realistic images, cloned voices and videos depicting people saying or doing things that never happened. Such content can spread rapidly through social media and messaging platforms, making it difficult for the public to verify its authenticity before it influences public opinion.
SynthID Detector offers journalists, election observers and voters an additional way to examine suspicious campaign footage, purported political speeches and questionable audio recordings.
However, the tool is not a comprehensive solution to election-related disinformation. Its effectiveness depends on whether the content contains a watermark that the detector can recognise.
Detector Cannot Identify Every AI-Generated File
SynthID Detector is not a universal AI-content detector. It primarily identifies watermarks associated with Google and participating technology companies, meaning content generated by unsupported systems or without a compatible watermark may go undetected.
The absence of a SynthID watermark does not prove that a recording is authentic or entirely human-created.
The tool also cannot independently establish whether the events shown in a genuine recording have been accurately described. Authentic footage can still be shared with a false caption, date or location.
Users are therefore advised to combine the detector’s findings with other verification methods, including checking the original source, examining available metadata, conducting reverse-image searches and contacting people or organisations connected to the content.
Google has also integrated media-verification features into products such as Search, Gemini and Chrome. The company said these features collectively handle more than one million verification requests daily.
The wider release of SynthID Detector reflects growing efforts by technology companies to improve transparency around AI-generated media. For Nigerian newsrooms and voters, it provides another verification resource ahead of the 2027 elections, although conventional fact-checking remains essential.
