We got tired of not knowing what was true
EnoughBullshit started from a simple frustration: every day we read dozens of claims on social media, news sites, and messaging apps, and we had no quick way to know which ones were real.
What it does
You paste any text — a tweet, a newspaper excerpt, a WhatsApp message, a political speech. EnoughBullshit reads it, identifies every verifiable factual claim inside it, and runs each one through an AI fact-check.
Within seconds you get a list of claims, each with a verdict: true, mostly true, misleading, missing context, unverified, mostly false, or false. No fluff, no opinion, just the call.
Instant results
Preliminary verdicts in under 10 seconds, powered by Groq's fast inference
5 languages
Works in English, Italian, French, German and Spanish. Detects the language automatically
AI fact-check
Uses llama-3.3-70b to reason about factual accuracy based on its training knowledge
Real news sources
Cross-references claims against GDELT's live global news feed to upgrade preliminary verdicts with real evidence
How the verdict works
Every claim gets two passes. First, the AI checks it against what it already knows from its training data and gives a preliminary verdict. This happens instantly and covers most common claims accurately.
In the background, we also search GDELT — a free real-time database of global news coverage — and pull in relevant articles. If we find solid evidence, the preliminary verdict gets replaced with a source-backed one. This whole process takes under a minute and happens silently.
Limitations worth knowing
EnoughBullshit is a tool, not a journalist. The AI can be wrong, especially on recent events or highly specialized topics. Verdicts marked with the "AI knowledge" badge mean no external sources were checked — treat those as starting points, not conclusions. For anything consequential — medical, legal, financial — always verify with a professional.
Why the name
Because we mean it. There's too much noise, too many half-truths, too many viral lies that never get corrected. The name is direct because the problem is too.
The tech
Built with Next.js, TypeScript and SQLite. The AI runs on Groq's inference API using Meta's Llama 3.3 70B model. News search uses the GDELT Project's free API. Everything is open, no paywalled data sources.
We don't use tracking analytics, we don't sell data, we don't run ads. The tool exists because it's useful.