Slop or Not for Mac: On-Device AI Slop Detector App

The AI slop hits your Mac long before it hits your phone. It lands as a supplier email in Mail, a student draft in Pages, a blog hero image dropped into a Figma board, a press photo pulled into a CMS. Slop Or Not now runs natively on Mac, so you can flag AI-generated text and images on the device where the content actually arrives.

Slop Or Not for Mac is an on-device AI slop detector for the Mac App Store. It checks images for C2PA Content Credentials (used by OpenAI), IPTC photo metadata (used by Google and many others), and runs an on-device AI image detection model called OmniAID when the metadata is missing. It also detects AI-generated text using a custom on-device transformer. Everything runs on the Apple Neural Engine through CoreML, and nothing leaves the Mac.

This post covers what the Mac app does, how the detection works, what the free tier gets you, and how the premium plans are priced. It does not cover the iPhone-only features.

Slop Or Not for Mac flagging an AI-generated cat astronaut image as Certainly AI Slop with a Google source badge — Slop Or Not for Mac reading a Google IPTC source credential and flagging the image as Certainly AI Slop

Key Takeaways

Slop Or Not is available on the Mac App Store as a native app for Mac with Apple silicon.
AI image detection reads C2PA Content Credentials (OpenAI), IPTC metadata (Google and many others), and falls back to OmniAID, an on-device AI image detection model.
AI text detection uses a custom on-device transformer trained on millions of human and AI samples.
Detection runs on the Apple Neural Engine through CoreML. Text and images stay on the Mac, and there is no account to create.
Free tier gives 3 checks a day, or 5 with a streak. Premium plans cost $2.99 a month, $19.99 a year, or $49.99 lifetime.

Slop or Not for Mac Is Now on the Mac App Store

You can download Slop Or Not for Mac on any Mac with Apple silicon. The app ships as a native Mac build, not a window-shaped iPad port. The interface uses the standard Mac toolbar and menu surface, and you can drag an image straight from Finder, Photos, or Safari into the check window.

Two workflows cover most of what people do on the Mac:

Paste-and-check for text. Paste an essay, a press release, or a long product description. Slop Or Not analyzes it on-device and returns a verdict. There is no character limit and no daily token budget for the analysis itself.
Drag-and-drop for images. Drop a photo into the window. Slop Or Not reads the embedded credentials first, then runs the on-device image detector if no credentials are present.

Reading C2PA and IPTC Credentials on Mac

C2PA Content Credentials and IPTC photo metadata are the two industry standards for marking an image as AI-generated. Slop Or Not for Mac reads both and uses them as the primary detection signal whenever they exist.

C2PA stands for the Coalition for Content Provenance and Authenticity. OpenAI ships C2PA on every image generated by ChatGPT and DALL-E. Adobe Firefly and Microsoft Designer ship it too. The credential is a signed manifest embedded in the file that records what tool produced the image, when, and what edits were applied later.

IPTC photo metadata is older and more widely supported. Google's image generators write IPTC DigitalSourceType values to mark synthetic media. Many camera and editing tools read and respect those fields.

When either credential is present and intact, Slop Or Not reports a confident verdict. The brand's public accuracy figures put detection of watermarked C2PA and IPTC images at 100 percent in internal tests, with the standard caveat that results can vary as new evasion methods emerge.

OmniAID Catches AI Images Without Watermarks

Watermarks get stripped. Screenshots, re-encodes, and aggressive social-media compression all destroy embedded credentials, and an attacker who wants to slip AI content past a check has every reason to remove them. That is what OmniAID is for.

OmniAID is the on-device AI image detection model Slop Or Not ships in the Mac app. The model is described in the OmniAID paper, which introduces a decoupled Mixture-of-Experts architecture that separates content-specific semantic flaws from content-agnostic visual artifacts. It runs locally on the Apple Neural Engine, looks at pixel-level signal that does not depend on metadata, and returns a verdict in seconds. It catches outputs from ChatGPT image generation, Google Gemini, Midjourney, and other modern generators when the embedded credentials have been wiped.

In internal tests, OmniAID reaches around 94 percent accuracy on AI-generated images that have no metadata to read. Accuracy drops as new generators appear and as humanizing tricks evolve, which is why the model updates over time. For a deeper look at the no-metadata case, see the post on detecting Nano Banana 2 images on iPhone — the same detection pipeline runs in the Mac app.

AI Text Detection on Mac With No Word Limits

AI text detection on the Mac uses a custom transformer that Slop Or Not trained in-house and ships as a CoreML model inside the app bundle. The model targets English text and is tuned for the kinds of writing people actually paste into a checker: essays, blog posts, emails, product descriptions, social posts.

Compared with a browser-based AI text detector, the Mac workflow has a few practical differences:

No word-count limits. Cloud detectors meter usage and ask you to chop a long document into pieces. Slop Or Not analyzes a full document in one pass.
No upload. A teacher checking a draft, a journalist checking a tip, or a lawyer checking a discovery document never sends the text to a third party.
No account. Open the app and run a check. There is nothing to sign in to.

In internal tests the text detector reaches around 95 percent accuracy. Results vary on text that has been deliberately humanized, which is why the app also includes a Text Cleanup utility that strips zero-width spaces, homoglyphs, and other hidden tricks before the check runs. There is more on the text workflow in the free ZeroGPT alternative and iPhone ChatGPT detector posts.

Why the AI Slop Detector Runs On-Device on Mac

On Mac, the AI slop detector runs entirely through the Apple Neural Engine using CoreML. The Neural Engine is the dedicated machine-learning accelerator built into every Apple silicon chip from the M1 onwards. CoreML is Apple's framework for executing machine-learning models against that accelerator from a sandboxed app.

That hardware path matters for two reasons. The first is speed, which is what makes paste-and-check usable as a workflow. The second is the privacy guarantee. The model weights ship inside the app bundle, the pixels and the text never leave the sandbox, and there is no server-side cache for a third party to compromise or subpoena.

That guarantee is the main thing the Mac build adds for the journalist, the teacher, the lawyer, and the medical writer. It is also the reason cloud detectors are not a substitute when the document is sensitive.

Pricing: Free, Monthly, Yearly, or Lifetime

Slop Or Not for Mac is freemium: the download is free, the detection models live in the app bundle, and the free tier gives:

3 checks a day by default, across text and images combined.
5 checks a day with a streak, which kicks in once you have used the app on consecutive days.

If you need more headroom, three premium plans lift the daily cap.

$2.99 a month for the rolling monthly plan.
$19.99 a year, which works out to about $1.67 a month for the same access.
$49.99 lifetime for a one-time purchase that covers the app for as long as it ships.

The lifetime plan suits power users and small teams who want a fixed cost. The lifetime plan also supports Apple's family sharing feature, one purchase can be used by up to 6 family members - making it ideal for families. There is no enterprise tier and no separate cloud subscription, because the detection itself runs on the Mac you already own.

Frequently Asked Questions

Does Slop Or Not for Mac run on Intel Macs? No. The detection models target the Apple Neural Engine on Apple silicon, so the Mac app requires an M-series chip (any Apple silicon Mac shipped from late 2020 onwards).

Does it run offline? Yes. After the initial download from the Mac App Store, every detection check runs against the local CoreML models. No network call is required to analyze text or images.

How accurate is it on Mac? Internal tests put detection of watermarked images (C2PA or IPTC) at 100 percent, OmniAID at around 94 percent on AI images without watermarks, and the text detector at around 95 percent. Accuracy can vary with new AI models and with humanizing tricks designed to evade detection.

Does Slop Or Not upload my files? No. Text and images stay inside the app sandbox on the Mac. There is no cloud processing step and no account, which is why Slop Or Not works for confidential documents and images that cannot leave your device. See the privacy policy for the full data-handling statement.

Can I share Premium between iPhone and Mac? Premium subscriptions are tied to the Apple ID that bought them and follow the standard Mac App Store and App Store family-sharing rules. If your iPhone and Mac share the same Apple ID, the same plan covers both.

Get the Mac Build

Download Slop Or Not for Mac from the Mac App Store and run your first check free. If you have not used the iPhone app, the ChatGPT detector for iPhone post covers the same workflow on the smaller screen, and the Forensically alternative post covers the image-detection trade-offs against a browser-based forensics tool.

The short version: install it, drag in a file, and stop guessing.