ChatGPT Watermark: What OpenAI's textGrain Means for Text
OpenAI announced textGrain, a ChatGPT watermark hidden in word choice. How it works, where it applies, who can detect it, and what it cannot prove.
Practical tips and tutorials for AI content detection across Apple devices. Learn to check for AI-generated text and images while keeping your workflow private.
Want to check something now? Use the free online AI detector or browse the AI detector tools for iPhone and Mac.
Comparing detector tools instead of researching from scratch? See our AI slop detector comparison hub for side-by-side guides.
OpenAI announced textGrain, a ChatGPT watermark hidden in word choice. How it works, where it applies, who can detect it, and what it cannot prove.
AI watermark removers delete hidden characters and file metadata. The mark inside the pixels and inside the word choices stays. We tested it.
Claude's text watermark changes word choices; supported Claude image files carry C2PA. See what each signal proves and why removal guarantees fail.
AI detector accuracy on a public benchmark: Slop or Not catches 99.5% of AI text at a 5% false-positive rate on RAID, adversarial attacks included.
Check a photo for AI provenance: read SynthID watermarks and C2PA Content Credentials first, then fall back to pixel-level AI detection.
A C2PA remover like NoC2PA strips C2PA and IPTC fields, but SynthID watermarks and pixel checks can still flag a cleaned AI image.
OpenAI and Gemini image verifiers are provider specific. Slop or Not checks SynthID, C2PA/IPTC metadata, and image signals when the source is unclear.
Use Slop or Not as an AI Detector MCP server so Claude, Codex, Cursor, Hermes Agent, and OpenClaw can check text and images locally on Mac.
Run the Agentic Humanizer in Cursor, Gemini CLI, or OpenCode and watch each rewrite move its Slop or Not score. No tool can guarantee a bypass of AI detectors.
Install the free Agentic Humanizer skill for Codex CLI and Claude Code to rewrite AI text in your own voice. Slop or Not can score each pass on your Mac.
Readability score checks in Slop Or Not run on-device in 7 languages, using Flesch in English and an established formula in each of the other six.
Slop or Not is on the Mac App Store. The free AI slop detector checks AI text and images on-device using SynthID, C2PA, IPTC, and an AI image detection model.
Slop Or Not checks Nano Banana 2 AI images with SynthID, metadata, and an on-device image model. iPhone and Mac checks stay private.
Free ChatGPT detector for iPhone with no word limits. Detect AI-generated text privately on-device with industry-best accuracy. No uploads required.
Three ways to check if an image came from ChatGPT: read C2PA Content Credentials, check SynthID watermark signals, then scan the pixels on-device.
Past the ZeroGPT character limit? Slop or Not checks any length on iPhone and Mac with no word cap per document or per month, and nothing leaves your device.
Help us understand how this site is used. Privacy policy