Verifying whether a video, image or audio clip has been manipulated is getting harder every year, as generative AI makes convincing fakes easier to produce than ever. vera.ai is a Horizon Europe research project (2022\u20132025) that set out to give journalists, fact-checkers and educators AI tools to keep pace, building on earlier EU-funded work through the InVID (2016\u20132018) and WeVerify (2018\u20132021) projects.<\/p>\n\n\n\n
The project’s main output is a free browser extension, the “Fake News Debunker” plugin, which bundles around 20 verification tools into one toolbar. Established features include reverse image and video search across seven engines, a keyframe tool that breaks a video into stills for easier searching, metadata and copyright checks, and an OCR tool that extracts and translates text embedded in images. vera.ai added newer AI-driven modules on top of these: a synthetic image detector, a deepfake video detector, and a voice-clone detector for AI-generated audio. A built-in Classroom section offers ready-made examples of manipulated media with explanations, which makes it a genuinely useful teaching aid for media literacy sessions, not just a professional tool.<\/p>\n\n\n\n
The plugin has real credentials: it won first prize in the US State Department’s 2021 Paris Tech Challenge, and the Poynter Institute has called it “one of the most powerful tools for spotting misinformation online.” It’s used by around 100,000 people a month, including teams at AFP and Bellingcat. That said, the newer AI-detection features are still labelled experimental by their own developers, and are best treated as a strong first signal rather than a definitive verdict. A few of the more advanced tools also require free registration as a journalist, fact-checker or researcher, which may limit access for casual classroom use. Worth noting too: most modules run locally or query external search engines, but the deepfake and voice-clone tabs upload material to project servers for analysis, so it’s not the tool to reach for with sensitive or source-identifying content.<\/p>\n\n\n\n
Overall, it’s a well-evidenced, actively maintained resource (last updated July 2026) that sits squarely at the intersection of AI literacy and media literacy \u2014 two areas central to MLA’s work.<\/p>\n\n\n\n