How does Slop Squash tell if an image is AI-generated?
It layers several checks. Provenance first: Content Credentials (C2PA) manifests are parsed and their signatures verified against the C2PA trust list, and the EXIF and IPTC fields are read for generator names and the “trained algorithmic media” source type. Then watermarks, where a public verifier exists. When a platform has stripped every byte of metadata, a pixel classifier can still attribute the image to a generator such as Nano Banana, Imagen, GPT or Midjourney. A trusted signature that says the image was generated pushes the score to at least 90%. The full list is on the detectors page.
Can it detect images made by Gemini or ChatGPT?
Often, yes. Every ChatGPT image since May 2026 carries a SynthID watermark and a Content Credentials manifest, and OpenAI runs a public verifier, so those are checked directly. Google signs Gemini, Imagen and Nano Banana output with Content Credentials, which Slop Squash validates cryptographically when the platform preserved them. Google’s own SynthID watermark can only be read by Google, so when a site has stripped the metadata the generator classifier gives a probability and a best guess at the source instead.
Is Slop Squash a SynthID checker?
Partly. SynthID is verified wherever a verifier is open to third parties: OpenAI’s provenance endpoint reports SynthID and Content Credentials signals for OpenAI-made images. Google’s SynthID for images is an undisclosed learned watermark with no verifier outside Google, and SynthID for text is keyed to the generator, so neither can be decoded by an extension or by anyone else. Slop Squash says so in the flag rather than guessing, and relies on Content Credentials and classifiers for Google-generated images.
What happens on Reddit, X and YouTube?
The extension has adapters for X, Reddit, YouTube comments, Facebook, Instagram, LinkedIn, Threads, Bluesky and Hacker News, plus a generic one for other sites. As you scroll, the posts, comments and images near the viewport are checked. When something scores above your threshold, a small glass circle appears in the corner of the post. Tap it and a card explains the evidence: which detectors ran, what they found and how confident the result is.
Can it detect AI-written text reliably?
No tool can, and Slop Squash does not claim to. Text watermarks are keyed to the generator and cannot be read by anyone else, so writing is scored on patterns: sentence-length rhythm, vocabulary that models overuse, lists of three, em-dash density, markdown headings inside comments and a suspicious absence of typos. Every triggered pattern is listed in the card. The verdict is AI at 70% or more and uncertain between 40% and 70%, and the circle only appears above your own threshold, which you can move in settings. Treat a text flag as a best guess.
What does the extension send off my device?
On the free tier: the text of the posts and comments it checks, the URLs of the images it checks, and the hostname of the page (never the full URL) for site statistics. Nothing about your account or browsing history. Our API fetches an image by its URL to read the metadata and may pass the image file, and nothing else, to a provenance verifier. Turn on “On-device only” in the popup and nothing leaves the browser: detection is limited to the writing-pattern heuristics and the Content Credentials it can read in the page. The privacy policy has the exact list.
Can it hide AI-generated posts automatically?
With Pro, yes. Auto-block turns the flag into an ad blocker for slop: anything scoring as AI collapses to a one-line pill that reads “AI content hidden” with a show button, so you never see it unless you ask. There is a per-site toggle, per-site thresholds, “always hide this account” rules and synced block lists, plus an optional recheck that asks a generator classifier and a vision-language model for a second opinion. Rechecks are cached, so a post one Pro user checks is known to everyone.
What does Pro cost, and what does lifetime mean?
Pro is $3 a month or $39 once. Lifetime means one payment for Pro for the life of the product, including future Pro features, with no renewals. Monthly renews until you cancel from your account page, and you keep Pro until the end of the period you paid for. The free tier needs no account and does not expire. Details are on the pricing page.
Can I get a refund?
Your first Pro purchase, monthly or lifetime, is covered by a 14-day money-back guarantee: email support@slopsquash.com within 14 days of the payment and it is refunded in full. Later payments are refunded at our discretion, and nothing here limits the refund rights you have under consumer law where you live. When a refund is issued, Pro ends immediately.
How accurate is it?
Accurate enough to be useful, not enough to be a verdict. A verified signature or watermark is near-certain when it fires; classifiers and writing patterns are probabilities, and the card always shows which one produced the flag. For comparison, in a 2025 Royal Society Open Science study typical adults picked the AI-generated face in a pair only 30% of the time, and trained super-recognisers managed 41%, both below chance. A tool that reads the metadata people cannot see, and shows its working, beats guessing.
What if a flag is wrong?
Open the card and choose “Not AI”. Reports go into a community tally keyed by a hash of the content, with nothing recorded about who reported it, and enough reports move the score for everyone. You can also raise your threshold, turn the extension off for that site, or email support@slopsquash.com with a link to the post.
Is there a creator program?
Yes. Creators get a link; anyone who follows it gets 30% off their first month of Pro monthly, and the creator gets 30% of what they actually pay. Payouts run weekly through Stripe once you reach $20. Apply on the creators page.