The Uncensored AI Fact Checker
Most fact-checking starts with someone deciding which claims deserve attention. BSmeter skips that step entirely: paste any YouTube, TikTok, Instagram Reel, or podcast URL and its AI scores every verifiable claim it finds, on any topic, explains each verdict, and searches for evidence sources on the weakest claim (the 3 weakest on Basic and Premium) so you can inspect them yourself. Prefer it blunt? Roast mode runs on an uncensored AI model that says exactly what it sees.
What you get
- No Topic Off-Limits: Politics, health, science, finance, viral conspiracy clips: BSmeter analyzes whatever you give it. There is no editorial desk deciding which subjects are worth checking and which get waved through. If a video makes verifiable claims, its main ones get scored.
- Roast Mode, Genuinely Uncensored: Roast mode runs on an uncensored AI model that delivers brutally honest, joke-first commentary on what a video actually claims. It is unfiltered in tone but built on the same claim-by-claim analysis as the standard report, so the punchlines land where the scores point.
- Every Verdict Shows Its Work: Unfiltered does not mean unaccountable. Each claim gets a 0-100 credibility score with a plain-language explanation, and for the weakest claim (the 3 weakest on Basic and Premium) BSmeter searches for evidence sources you can open, so you can judge the reasoning yourself instead of trusting a rating on faith.
- Fallacies and Framing Flagged: Beyond scoring claims, BSmeter names the logical fallacies in the argument and the rhetorical moves around it, like Missing Context, Cherry-Picked data, or Anonymous Authority. Spin gets called out even when every individual fact technically checks out.
- Checks As You Watch: The Chrome and Firefox extension checks the video you are already watching on demand, and pasted URLs work the same way, with results usually within minutes. You find out whether a claim holds up while the video is still circulating, not days after the moment has passed.
Frequently asked questions
- What does "uncensored" mean in an AI fact checker?
- It means no editorial gatekeeping over what gets checked. Traditional fact-checking teams select and curate which claims to review, which is reasonable given limited human hours, but it means most video claims are never examined at all. BSmeter analyzes any video you give it on any topic, extracts its main verifiable claims, and scores each one 0-100, searching for evidence sources on the weakest claim (the 3 weakest on Basic and Premium). Uncensored refers to coverage and tone, not standards: the same non-partisan rubric applies to every claim. Results are AI-generated and can be wrong, so check anything that matters.
- Is Roast mode really uncensored, or is it just marketing?
- Roast mode runs on an uncensored AI model, so the commentary is brutally honest and joke-first rather than hedged and polite. What keeps it from being empty snark is that the roast is built on the same claim analysis as the standard report: transcription, claim extraction, 0-100 credibility scores, and evidence sources for the weakest claim (the 3 weakest on Basic and Premium). It says what it sees, and you can check what it saw.
- Can I fact-check controversial or political videos with BSmeter?
- Yes. Politics, health claims, scientific disputes, finance gurus, and viral outrage clips all go through the same pipeline: transcribe, extract the verifiable claims, score each 0-100, search for evidence sources on the weakest one (the 3 weakest on Basic and Premium), and flag logical fallacies and misleading framing. BSmeter is non-partisan by design, since the scoring works the same regardless of which side a claim helps, and each verdict comes with an explanation you can double-check on your own.
- Is there a free uncensored fact checker I can try?
- BSmeter's free tier includes three analyses per day at bsmeter.ai, with no topic restrictions. Basic is $3.99/month for 15 analyses a day, and Premium is $8.99/month and handles videos up to two hours. Both paid plans are month-to-month — cancel anytime.