Slop or Not
Use it for people who want native Apple checks or the free online checker for text, images, and provenance clues.
vsGSAI Image Detector
Slop or Not vs Google SynthID: An AI Image Detector for What Nobody Watermarked
Use both. Google's tool is the one for watermarked images, video, and audio from Google and its partners. Slop or Not is the one for the image nobody watermarked and for text, and in the iPhone, iPad, and Mac app the check never leaves your device.
SynthID tells you if a big company watermarked it. Scammers pick the generators that don't. On October 7, 2026, Google opened its SynthID Detector to everyone. At synthid.com it checks images, video, and audio for the Google SynthID watermark that Google and partners including OpenAI, NVIDIA, Kakao, and soon Apple put into what their AI makes, and Google has built SynthID checks into Search, Chrome, and the Gemini app. Slop or Not reads the same watermark in Google and OpenAI images. But a watermark exists only where a company chose to put one. Open-weight models run on someone's own computer carry no SynthID and usually no signed C2PA credentials, and a screenshot strips whatever metadata their apps write. Slop or Not also verifies C2PA Content Credentials, reads generator fingerprints, and scans the pixels, so generators that watermark nothing still get a result.
Privacy, Pricing, and Workflow Matrix
Public product, pricing, documentation, and benchmark pages can change. Source links are listed below the table so the cells stay readable.
| Feature | Slop or Not | GSGoogle SynthID |
|---|---|---|
| Best fit | On-device iPhone and Mac checks or free browser checks for text and images | Checking whether an image, video, or audio file carries the SynthID watermark from Google or a partner |
| Text detection | Checks writing in the native app and online checker | Not listed: the announcement covers image, video, and audio files |
| Image detection | Checks photos, synthetic images, and provenance clues in the app or online checker | Checks images for the SynthID watermark from Google and partners, including OpenAI Finds watermarked images only |
| Deepfake detection | Covers deepfake and synthetic-image checks as image detection | Checks video and audio for the SynthID watermark from Google and partners Finds watermarked media only |
| Metadata and provenance signals | Reads SynthID, C2PA, and IPTC signals in native and online image checks when a file carries them | Reads the SynthID watermark in images, video, and audio |
| Manual forensic tools | Reports detector and provenance evidence in native and online checks; not a manual forensic editor | The announcement describes no manual forensic tools |
| Evidence level | Probability verdict plus provenance signals, with internal-test caveats instead of 99.9% certainty claims | Reports whether a file was made or edited with AI from Google or a SynthID partner |
| Browser checker | Free online checker runs in any browser on Numen's private Mac server, not a third-party AI provider's | synthid.com checks Google and partner marks; Google has also built SynthID checks into Search, Chrome, and the Gemini app |
| Native Apple app | Native iPhone and Mac app for on-device checks; browser users can use the free online checker | Built into the Gemini app, per the announcement |
| API workflow | Pro Mac app includes a local CLI and MCP server; no public hosted API | The announcement describes no public detection API |
| Classroom workflow | Useful for checking student writing on-device in the app or through the free online checker | The announcement describes no classroom workflow |
| Free tier & limits | Free daily checks in the iPhone and Mac app and a free online checker in any browser. Pro removes the native daily cap; the native app has no word cap per document or per month. | Open to everyone, globally in English; the announcement states no pricing or limits |
| Privacy | Native checks stay on the iPhone or Mac and never send submitted text or images to a third-party AI provider. Online checks run on Numen's own Apple hardware, never a third-party AI provider's. | The announcement does not describe how uploaded files are handled |
| Main limitation | On-device privacy requires iPhone or Mac; browser checks run on Numen's private Mac server, not on-device | Finds only media that Google or a partner watermarked with SynthID; the announcement describes no check for unwatermarked content |
✓ = supported, ✕ = not offered, – = not publicly documented. Details reviewed October 2026 from public product, pricing, documentation, and benchmark pages.
Sources reviewed
Where Slop or Not Fits
When an image arrives with no watermark and no credentials, a watermark check has nothing to read. Slop or Not still scores the pixels with its image model and returns a probability, and in the app that happens on your iPhone, iPad, or Mac.
- Slop or Not runs its own detection models, so no third-party AI provider sees your text or images. Native checks run on Apple silicon, accelerated by the Apple Neural Engine; online checks run on Numen's self-hosted Apple hardware.
- Free daily checks in the iPhone and Mac app and in the free online checker. Pro removes the daily cap, and native checks have no word cap per document or per month.
How They Compare by Workflow
Slop or Not is not the right answer for every workflow - pick the tool whose strongest workflow matches the check you need to run.
GS
Google SynthID
Use it for checking whether an image, video, or audio file carries the SynthID watermark from Google or one of its partners.
Common Questions
Does SynthID detect ChatGPT and OpenAI images?
Yes, when the mark is there. Older images, and images from some products or export paths, may carry none. OpenAI is one of Google's SynthID partners, and Google's checker reads its watermark. Slop or Not reads it too, with its own calibrated threshold for OpenAI images. In the iPhone, iPad, and Mac app, every hit links to Google's and OpenAI's own verifiers so you can check the result. A current ChatGPT image usually also carries C2PA Content Credentials until a screenshot or an upload strips them. The SynthID mark lives in the pixels, so it usually survives that.
What is the difference between SynthID and C2PA?
SynthID is an invisible watermark woven into the pixels, which is why it usually survives screenshots and re-encodes. C2PA Content Credentials are signed metadata attached to the file. They can say which tool made or edited an image, but a screenshot or a metadata strip removes them, and the official C2PA trust list admits only products that pass its conformance program. Slop or Not reads both: SynthID in Google and OpenAI images, and C2PA credentials verified against a bundled trust list, so a forged credential fails.
Can you remove SynthID from an image?
Heavy edits can weaken a mark until a check comes back inconclusive. But removing a mark doesn't remove the AI from the pixels. When no SynthID mark or credential is readable, Slop or Not's image model still scores the pixels and returns a probability that the image is AI-generated. A missing mark is never proof that a person made the picture.
Does SynthID work on text?
There is a text version, SynthID Text, and Google has open-sourced it for developers. Google's new checker is announced for images, video, and audio. Slop or Not never reads SynthID Text. Its text check is an AI text detection model, the same one listed on RAID, the public AI-detection benchmark, where it catches 99.5% of AI text at a 5% false-positive rate with adversarial attacks included. It reads English text whether or not anyone watermarked it.
Is synthid.com Google's official tool?
Yes. Google's own announcement on October 7, 2026 names synthid.com as the place to check a file, alongside the checks built into Search, Chrome, and the Gemini app. It launched globally in English.
Privacy tools from the same team
More From Numen
AI can now fake a profile photo or wrap a scam in a link that looks safe to tap. Numen builds the apps that put you back in charge: spot AI-generated text and images with Slop or Not, run the best open-source models uncensored and private with Private LLM, strip trackers from links and QR codes before they follow you with Clean Links, and generate images from frontier models on your own hardware with Private Diffusion.
One belief runs through all four: your data belongs to you, and no one else.
Private LLM
Run local AI models privately on your iPhone or Mac, even when you are offline.
Clean Links
Remove trackers from links and check QR codes before a tap follows you somewhere else.
Private Diffusion
Run the latest frontier image models on your iPhone, iPad, and Mac. Private Diffusion trains each one with quantization aware distillation (QAD), so the small model keeps the quality of the full size one.
Related Comparisons
Keep comparing adjacent tools before you pick a workflow.
Slop or Not vs GPTZero
GPTZero lives in Canvas and Google Docs and never leaves the browser — and it never checks an image, despite the 'AI Vision' name.
Slop or Not vs ZeroGPT
ZeroGPT needs no account and returns a verdict in seconds, but paste in more than 15,000 characters and it stops reading.
Slop or Not vs Originality.ai
Originality.ai tracks every scan in a dashboard built for agencies, but there's no free tier — every check burns a credit at roughly 100 words apiece.
Slop or Not vs Copyleaks
Copyleaks plugs into seven different LMS platforms and even flags deepfake video, but the sentence-level detail and full reporting sit behind an enterprise account.
Slop or Not vs Winston AI
Winston AI bundles plagiarism checks and writing feedback with its AI verdict, but the free lite checker stops at 2,000 characters and the real trial runs out in 14 days.
Slop or Not vs Turnitin
Turnitin's AI-writing indicator is built into a workflow schools already trust, but there's no public version — access runs through a school subscription or not at all.