Refract

Refract analyses any news article for political bias, funding transparency, editorial purpose and factual accuracy — before you finish reading it. Two ways to check a story Quick Scan gives you a trust score in about fifteen seconds. Three independent AI models — Claude, GPT-4 and Gemini — each score the article for bias, purpose, funding and source credibility. Refract shows you the average, and tells you clearly whenever the models disagree. Deep Verify goes further. It pulls out every factual claim in the article and searches for the primary source behind it — a government dataset, a court record, an official report. You get a verdict on whether each claim holds up, with a link to what we found. Why three models, not one A single AI carries a single set of blind spots. Refract never relies on just one model's judgement. Where Claude, GPT-4 and Gemini agree, you can trust that's a real signal. Where they don't, we say so — some stories are genuinely contested, and pretending otherwise isn't honesty. What Refract checks • Political lean and emotional language • Editorial purpose — is this reporting, persuading, or provoking? • Factual accuracy, with verified and questionable claims counted • Source credibility and publication track record • Funding, ownership and potential conflicts of interest • Narrative originality — is this story echoing a talking point? Built for anyone who reads the news Whether you're trying to spot bias in political coverage, checking a viral claim before you share it, or teaching media literacy, Refract gives you a second opinion — from three AI models at once, not one. Free to start. 10 credits a month, no card required.