scan
open the popup and hit scan. verif·ai reads the current tab on demand — images and text. nothing runs in the background, nothing leaves your machine.
free · chrome extensionspot ai content,
spot ai content,
keep what’s real.
verif·ai highlights ai-generated images and text on the page you’re reading.
one click. an honest read on every image and text block — credentials when they exist, a score when they don’t. no accounts, no servers, no history.
free · on-device · part of ogbuilds
theblog.example/the-future-of-design
ai · credentials
ai text · 78%
- content types scanned
- 2content types scanned
- accounts required
- 0accounts required
- on-device, on-demand
- 100%on-device, on-demand
- click to scan
- 1click to scan
how it works
three calm steps, all on demand
detect
images are checked for embedded content credentials — the c2pa / iptc standard that dall·e, firefly and gemini stamp into their output. text gets a transparent on-device heuristic score.
see
likely ai content is highlighted right on the page — a dashed outline on images, a soft tint on text — and summarised in the popup.
built for trust
the web is filling up with ai-generated images and text. verif·ai gives you a fast, honest read on what you’re actually looking at — without the doom.
theblog.example/the-future-of-design
ai · credentials
ai text · 78%
verif·aion2 ai images · 1 ai text block
this page- imagehero illustrationai ✓credentials
- textintro paragraph78%medium
- imageinline figureai ✓credentials
content credentials + on-device heuristic · part of ogbuilds
theblog.example/the-future-of-design
ai · credentials
ai text · 78%
on the page
highlights you can trust
no dashboards, no exports, no busywork. verif·ai marks suspected ai content exactly where you're reading it — a dashed outline around images and a calm tint behind text blocks.
click any item in the popup to jump straight to it on the page.
scan results
- imagedall·e illustrationai ✓
content credentials: ai-generated
- textopening paragraphmedium71%
- imagestock photounverified
no embedded credentials — not proof it’s real
a real signal
a score you can read
for images, verif·ai reads the provenance credentials baked into the file — an authoritative “this was made by ai” signal, not a guess. for text, you get a 0–100 score with a high / medium / low confidence level.
and it stays honest: when an image carries no credentials, verif·ai says exactly that — it never pretends “no signal” means “human.”
running locally
- accountsnone
- sign-innever
- server-side historynone
- scanson demand only
- your datastays on device
your machine, your call
private by design
no accounts, no sign-in, no server-side history. scans run on demand against the active tab and nothing else.
an enable / disable toggle puts you in control, always.
questions, answered
what is verif·ai?
verif·ai is a chrome extension that detects and highlights ai-generated content — images and text — on the web page you're currently viewing. you click scan, and it marks likely ai content directly on the page with a confidence score.
how does verif·ai detect ai-generated content?
for images, verif·ai reads the content credentials embedded in the file — the c2pa / iptc “digital source type” standard that dall·e, adobe firefly and google gemini stamp into their output. when an image is credentialed as ai-generated, that's an authoritative signal, not a guess. for text, verif·ai applies a transparent on-device heuristic with a 0–100 score and a confidence level.
does verif·ai send my browsing data anywhere?
no. the free extension needs no account and keeps no server-side history — scans run on demand against the current tab only, and your settings live in local browser storage. pro's optional hosted deep-scan sends only the flagged image's url for analysis, never the page you're reading.
which browsers does verif·ai support?
the mvp targets google chrome (and chromium browsers like edge, brave and arc) using manifest v3. firefox and safari support are planned for later.
can verif·ai detect ai-generated text as well as images?
yes — both. images are flagged when they carry embedded ai credentials (for example from dall·e, firefly or gemini); text is flagged by an on-device heuristic. each is reported separately in the popup. images whose metadata has been stripped — screenshots, many social re-uploads, or tools that embed nothing like midjourney — won't carry credentials, so they're reported as “unverified,” never as “human.” pro's hosted deep-scan covers exactly those images.
what are content credentials, and what about screenshots?
content credentials (the c2pa / iptc standard) are provenance data some ai tools embed directly in an image file, which verif·ai reads on-device. the catch: screenshots, file conversions and many social uploads strip this data, and some generators don't add it at all. so a missing credential never means an image is real — it just means there's nothing to read. covering those cases is exactly what pro's hosted deep-scan is for.
is verif·ai free?
the core extension is free and runs entirely on your device — no account, no tracking. pro ($1.99/wk, $5/mo or $39/yr) adds the hosted deep-scan for uncredentialed images, auto-scan as you browse and the toolbar badge. built in public as part of ogbuilds.
clean up what you read.
add verif·ai to chrome and get a calm, confident read on ai-generated content — everywhere you browse. free, and private by design.
no card required · no account · no tracking