Hey Product Hunt! 👋 I'm Pablo, creator of SlopTotal. Single AI detectors are notoriously unreliable—they hallucinate false positives, overfit on modern web copy, or break entirely when given historical prose. I built SlopTotal to bring transparency to AI text detection, operating like VirusTotal for AI content. Instead of a black-box percentage, it runs 23 independent detection engines locally in parallel on your own hardware and streams every engine's vote in real time. What makes it different: ⚡ 100% Local & Self-Hosted: Runs on standard CPU hardware (~4GB–8GB RAM) via Docker with zero third-party API dependencies or data leakage. 📚 Audited Against Historical Prose: We benchmarked the ensemble against Project Gutenberg literature (1532–1915) to eliminate bias against complex or archaic writing (Machiavelli scored 62.5% AI on single detectors before ensemble calibration!). 🛠️ Exposes Broken Weights: Shows every engine's vote so you can see when models disagree or fail. 🔍 Site-Check Feature: Includes an automated check that identifies footprint markers left by AI app builders like Lovable, v0, Bolt, Base44, and Replit. 🌐 Try the Live Demo: https://sloptotal.com ⭐ GitHub Repo (MIT License): https://github.com/pablocaeg/slo... I'd love your feedback, edge cases you test against it, or suggestions for new engines to integrate!

SlopTotal

VirusTotal for AI text. 23 engines, self-hosted on CPU.

VirusTotal for AI text. 23 engines, self-hosted on CPU.