Slop or Not: AI Slop Detector for Text and Images
Google's own detector checks only Google's images. OpenAI's checks only OpenAI's. Slop or Not is the only AI image detector that reads Google DeepMind's SynthID watermark from both Google and OpenAI images. Its AI text detection model caught 99.5% of AI text at a 5% false-positive rate on RAID, a public benchmark we don't run. Both checks happen on your iPhone or Mac, with no uploads and no word limits, per document or per month.
Our detection models are trained on millions of human and AI samples and run entirely on Apple silicon, accelerated by the Apple Neural Engine, never on a server. They flag content from ChatGPT, Gemini, DALL-E, Midjourney, and more.
Google's own detector checks only Google's images. OpenAI's checks only OpenAI's. Slop or Not is the only AI image detector that reads Google DeepMind's SynthID watermark from both Google and OpenAI images. Its AI text detection model caught 99.5% of AI text at a 5% false-positive rate on RAID, a public benchmark we don't run. Paste English text or pick a photo and try either check in your browser. It's free and there's no account to create.
Online checks cover English text and photos. They run on our own fleet of Macs, using models trained on millions of human and AI samples. We never send your text or photo to another company, and we delete it after processing. The image models flag output from DALL-E, Midjourney, Gemini, and more.
AI Image Detector: Spot Deepfakes and AI Photos on iPhone and Mac
Slop or Not is the only AI image detector that reads Google DeepMind's SynthID watermark from both Google and OpenAI images. Google's own detector checks only Google's images. OpenAI's checks only OpenAI's. Slop or Not reads both. That mark lives in the pixels, not in a metadata field, so it survives screenshots, re-encodes, and metadata stripping, and each provider gets its own calibrated threshold. When a mark is there, the result can name the provider. When there is none, you get a probability from the image model instead, because a missing watermark proves nothing.
AI Text Detector With Unlimited Word Count
On RAID, the public AI-text-detection benchmark, Slop or Not catches 99.5% of AI text at a 5% false-positive rate, adversarial attacks included. RAID's own maintainers scored that, on a leaderboard anyone can read. A public exam, not self-graded homework. See the full RAID results. The same model runs on your iPhone and Mac, so you can paste a whole essay with no word cap per document and no monthly word allowance, and the writing never leaves the device. The free app has a daily check limit; Pro removes it. Text detection works in English today, with more languages planned.
How Slop or Not Catches AI Slop
Every detection layer runs on your device. Nothing is uploaded or sent to an API, so there is no server queue.
99.5%
of AI text caught at a 5% false-positive rate, scored by RAID
3
layers an AI image has to get past: SynthID watermarks from both Google and OpenAI images, C2PA and IPTC credentials from many more generators, and our own model
0
uploads in the iPhone and Mac apps. Every check runs on your device
The SynthID watermark lives in the pixels, C2PA and IPTC credentials in the file
Many AI image generators leave a record of themselves in the file. Slop or Not reads two open credential standards: C2PA, which OpenAI, Microsoft, and Adobe sign, and IPTC, which Google and most major AI image tools write as plain text. It checks a signed credential against a bundled trust list, so a forgery fails instead of passing as proof. That read is a lookup, not a guess. SynthID works differently. That watermark is written into the pixels rather than a metadata field, so it survives the screenshot that strips a credential, and Slop or Not reads it from both Google and OpenAI images.
Try the SynthID detector in your browserImage model runs on Apple silicon, accelerated by the Apple Neural Engine
When no watermark or credential is present, our image model scans the pixels for subtle artifacts the eye misses. The whole pass runs on your device's Apple silicon, accelerated by the Neural Engine, never on a server.
Custom transformer for AI text detection
A custom-trained transformer checks every passage for AI fingerprints. It runs offline through CoreML on Apple silicon, accelerated by the Apple Neural Engine, with no word cap per document and no monthly word allowance.
You do not have to take our word for any of this. When a SynthID read names the provider, you can confirm a Google hit against Google's own verifier and an OpenAI hit against OpenAI's.
Image accuracy is based on our internal tests. Text detection is independently scored on RAID (99.5% detection at 5% false positives). Results can vary with new AI models and advanced methods designed to trick detectors.
Two Readability Scores in One Tap
Every text check includes readability scores, calculated on-device in seven languages. English uses Flesch Reading Ease and Flesch-Kincaid Grade Level. German uses Wiener Sachtextformel; Spanish uses Flesch-Szigriszt; Italian uses GULPEASE; Swedish, Danish, and Norwegian use LIX. Nine published readability formulae in all, five of them for English, computed with their published coefficients, so the same text returns the same numbers every time. Those scores tell you how hard a passage is to read. They do not tell you who wrote it, and they are only estimates. You do not need an account, and nothing is uploaded.
CLI and MCP Tools for AI Checks on Mac
Slop or Not now ships a Mac CLI and MCP server for Claude, Codex, Hermes Agent, OpenClaw, Cursor, and other AI tools. Detect AI text, check readability, clean hidden characters, detect images, get raw image model scores, and send the result back to your assistant. The CLI and MCP server are Pro features; the checks run on your Mac.
AI Slop Sticker Pack
Found a fake? Add an AI Slop sticker to an iMessage or social post. Sometimes a joke makes the point better.
How the AI Detector Works
Get the AI Detector
Download Slop or Not free from the App Store. The app installs a compact detection model that runs on Apple silicon, accelerated by the Apple Neural Engine. After setup, you can run checks offline, with no account needed.
Run an AI Check
Paste any length of English text or pick a photo from your photo library. Screenshots work too: Slop or Not reads the text inside the image and runs the same text check on it. The AI detector analyzes your text or photo on-device and returns a verdict in seconds, even without an internet connection.
Find AI Slop
See a clear verdict and a slop meter: the slop score from 0 to 100, next to the Human score on text or the Real score on a photo. If the text contains AI fingerprints, use Text Cleanup to strip hidden characters. It is a lookup table, not a rewrite: same input, same output, every time, and your words are untouched. Share your results directly from the app.
Who Uses an AI Slop Detector
Journalists & Fact-Checkers
Verify images and source quotes without uploading them. Checks run on your device instead of through OpenAI, Anthropic, or another cloud LLM. Sensitive material stays with you while you fact-check it.
Teachers & Students
Check a whole class set without splitting essays to fit a word cap or watching a monthly word balance drain. Teachers can flag AI-written essays, and students can review their own work on-device. Pro removes the daily check cap; classroom pricing is available in the app.
Online Dating
The FTC counted $1.14 billion in romance-scam losses for 2023, more than any other impostor fraud. Swindlers build fake dating profiles around faces no camera ever took. Run the photo through the image detector before you swipe.
Online Shoppers
Check product photos and customer reviews for AI-generated fakes. The AI text detector flags synthetic five-star reviews, and the image detector catches AI product shots.
Finance & HR Teams
Authenticate ID documents, résumés, and voiceover scripts. Reduce the risk of fraud and deepfake payroll scams.
Legal & Medical Professionals
Check contracts, briefs, or patient files for AI signals without sending them to the cloud. Slop or Not runs locally on iPhone or Mac, not through OpenAI or Anthropic.
Content Creators & Community Managers
Moderate user-generated posts for AI spam or disinformation. Keep forums, Discord servers, and social feeds trustworthy.
Social Media Users
Check viral posts and trending images before you share them. The AI detector spots AI-generated text and deepfake images in seconds, right on your device.
Looking for an AI Slop Detector Comparison?
Slop or Not is a private AI slop detector for iPhone and Mac. See how it stacks up against ZeroGPT, Forensically, Hive Moderation, Reality Defender, Proofademic, and FotoForensics.
Download the Free AI Slop Detector for iPhone and Mac
Run our compact detection model on-device to check text and images in seconds. After setup, you can run checks offline. The app is free and needs no account.
Free Online AI Detector for Text and Images
Give the detector an English paragraph or a photo, and it tells you how likely a machine made it. Checks run on our own Macs and are deleted after processing. No account needed.