Google opens AI watermark detector to curb Nigeria election deepfakes
Google has made its SynthID Detector platform publicly available, allowing anyone to upload images, videos or audio files to check for invisible AI-generated watermarks. The tool works with content created by Google, OpenAI, NVIDIA and Kakao systems, with Apple expected to join later. Access requires a Google, Apple or OpenAI account. The launch comes as concerns rise about AI-generated deepfakes, cloned voices and manipulated media ahead of Nigeria’s 2027 general elections, a threat flagged by the Independent National Electoral Commission. Journalists, fact-checkers and ordinary voters can now use the detector to verify suspicious campaign footage, purported speeches or fabricated endorsements before sharing or acting on them. After upload, the service scans for a recognised SynthID watermark and reports whether supported AI technology created or altered the file. Google notes the detector is not universal: it only identifies watermarks from participating companies, so a negative result does not prove authenticity. Users must combine it with other verification methods such as source checks, metadata analysis and reverse-image searches. Google says it has embedded SynthID in over 180 billion AI-generated images and videos and about 240,000 years of audio, with its other verification features handling more than one million media checks daily. For Nigerians facing increasingly sophisticated election-related misinformation, the platform offers an additional line of inquiry, though it does not replace traditional fact-checking. Will you use this tool to vet political content before sharing, or continue relying on established fact‑checking practices?