Google has made its artificial intelligence watermark-detection platform available to the public worldwide, offering journalists, fact-checkers and other users an additional tool to verify suspicious online content ahead of Nigeria’s 2027 general elections.
The company announced that its SynthID Detector is now accessible globally in English, allowing users to upload images, videos and audio files to check whether they contain invisible digital watermarks embedded by supported AI-generation systems.
According to Google, the platform can detect SynthID watermarks in content created using its own AI tools and those of participating technology companies, including OpenAI, NVIDIA and Kakao. Apple is expected to join the initiative later.
Users can access the service with a Google, Apple or OpenAI account.
Google said the tool was developed to help ordinary internet users, journalists and fact-checkers examine questionable media and obtain additional information about how certain content may have been created.
The public launch comes as political activities gather momentum ahead of Nigeria’s 2027 elections, raising concerns about the possible use of AI to spread misinformation, impersonate political candidates and influence public opinion.
With advances in AI technology, it has become increasingly easy to produce convincing photographs, cloned voices and videos showing people making statements or performing actions that never happened.
Such material can spread rapidly across social media and private messaging platforms, sometimes reaching thousands of people before journalists, public officials or fact-checkers can establish its authenticity.
The Independent National Electoral Commission (INEC) has identified the weaponisation of AI-generated disinformation and the volatility of social media as potential threats to the credibility and security of Nigeria’s electoral process.
Against this backdrop, Google’s new platform could provide Nigerian journalists and voters with another means of examining questionable campaign videos, purported speeches, fake endorsements and suspicious audio recordings.
Unlike visible labels or logos, SynthID embeds an invisible digital signal into AI-generated content during production. The signal is designed to be undetectable to the human eye or ear but recognisable by the company’s detection technology.
Google said the watermarks embedded in images and videos are designed to remain detectable even after common alterations such as cropping, filtering, compression and changes to video frame rates.
For audio files, the watermark is embedded in a way that cannot be heard by listeners and is designed to withstand modifications, including background noise, speed changes and MP3 compression.
When a user uploads a file, the detector scans it for a recognised watermark and reports whether it finds evidence that supported AI technology was used to generate or modify the content.
Google introduced SynthID in 2023 and subsequently tested an early version of its detection platform with journalists and media professionals before opening it to the wider public.
However, the company’s tool has limitations and cannot identify every piece of AI-generated content circulating online.
Its detection capabilities are largely restricted to watermarks embedded by Google and participating technology companies. Content produced with unsupported AI systems or without compatible watermarks may go undetected.
Consequently, a result showing that no SynthID watermark was found does not necessarily mean that the content is genuine or was created entirely by humans.
The tool also cannot independently confirm whether the events shown in a photograph or video actually occurred as described. Even genuine footage can be misleading when shared with a false date, location or caption.
Journalists and other users will therefore need to combine the technology with established fact-checking methods, including examining the original source, checking metadata, conducting reverse-image searches, comparing landmarks and contacting the individuals or organisations involved.
Google disclosed that more than 180 billion AI-generated images and videos had been watermarked with SynthID since the technology was introduced. It also said the system had been used to watermark approximately 240,000 years’ worth of generated audio.
The company’s other content-verification features are already available through Google Search, Gemini and Chrome, which Google said collectively handle more than one million media-verification requests daily.
Although SynthID also supports watermarking for text generated through certain Gemini products, the publicly announced detector is primarily presented as a verification tool for images, videos and audio.
Google said the initiative was part of its broader effort to give users more information about the content they encounter online.
For Nigeria, where election-related misinformation could become more sophisticated as the 2027 polls approach, the platform offers an additional way to investigate suspicious media before sharing or acting on it.
However, while the technology could help identify some AI-generated material, it is not a substitute for independent verification, responsible journalism and critical thinking.
