Hey everyone 👋 I built SynthGuard after watching creator friends get their accounts flagged or shadowbanned because AI detectors on Instagram, TikTok, OnlyFans and Fanvue kept catching their generated content — even when it was harmless promo material. Every existing "humanizer" had the same problem: you upload your stuff to their server, hope they don't store it, and pray it actually fools modern detectors. For NSFW creators and OFM agencies, that's a non-starter. So I went the opposite direction: everything runs locally in the browser. No upload, no backend processing, no logs of your media. The pipeline is 10 layers deep — PRNU sensor fingerprints, FFT disruption, real camera EXIF injection, chroma subsampling, double JPEG compression, and a few more passes that target the exact signals classifiers look for. The biggest evolution during the build: I originally focused only on images. But after talking to early users I realized the real pain is multi-format consistency — you generate an image with Midjourney, a caption with GPT, maybe a short clip with Sora, and all three need to pass detection. So I added a text humanizer and a video humanizer with the same privacy model. I also shipped a free AI image detector using 8 forensic signals (noise correlation, FFT energy, EXIF physics, block artifacts…) so you can verify your output with the same kind of math the platforms use — before you post. Free plan gives you 3 credits/week to try everything. Would love feedback on the detector accuracy and what platforms you're fighting most right now 🙏

SynthGuard.net

Make AI images, text & video undetectable — 100% in-browser

Make AI images, text & video undetectable — 100% in-browser