VeriVoice AI — Fact-Check Anything

VeriVoice AI is a fact-checking tool built around a simple observation: misinformation doesn't spread as text articles anymore, it spreads as WhatsApp voice notes, forwarded videos, and AI-generated clips — and every major fact-checking tool is still built for the old format. Users submit any audio clip, video, link, or headline. VeriVoice transcribes it (Speechmatics), searches live news sources in real time (Bright Data), and reasons only over that retrieved evidence — never guessing from training data. Within seconds, they get a traffic-light verdict, a confidence score, plain-language explanation with an action recommendation, and real cited sources they can click through and verify themselves. The verdict system distinguishes "confirmed," "disputed" (true but imprecise), "unsubstantiated," and "false" — rather than collapsing every uncertain claim into a single misleading label, which we consider essential for a tool people are trusting with real decisions. Because WhatsApp is a closed platform that public fact-checkers can't scan, we also built a WhatsApp Web browser extension: right-click any message in a real conversation and get a verdict inline, without forwarding it elsewhere first — a different approach from existing WhatsApp fact-check bots, which all require sending a claim out to a separate number. The whole experience is designed for non-technical and older users specifically, since they're both the most targeted by this kind of misinformation and the least served by dense, text-heavy fact-checking reports: a Senior-Friendly Verdict mode, audio readout of every verdict, and a downloadable shareable image so a debunked claim can be forwarded back into the same family group chat it came from.