The shortlist
5 things that do it.
Free web tools and a Chrome extension that check images and video for AI generation, and predict which generator made them.
- Free
- The Chrome extension is free with no login (Chrome Web Store listing).
- Paid
- price not checked
- Best at
- Free checks with no login, by right-click, upload or paste.
- The catch
- It shows a likelihood only. In NewsGuard's test Hive flagged none of 15 real photos but caught 9 of 15 heavily manipulated ones, and a Hive result was cited online to wrongly call a real video fake.
- chrome
- web
- api
Upload an image to see whether it is AI or real, with a generator breakdown, plus deepfake checks, mainly sold as an API.
- Free
- $5 in free credits to start, including 20 AI image checks.
- Paid
- from $5 per month (Pro, includes $10 of credits)
- Best at
- Naming the likely generator, and catching all 15 heavily manipulated images in NewsGuard's test.
- The catch
- In that same test it flagged 1 of 15 real photos as AI; its CEO said low image quality can sometimes cause a false positive. Its 98.9% accuracy figure is the vendor's own.
- web
- api
A drag-and-drop checker that reports how likely an image is AI-generated and whether faces were manipulated.
- Free
- The drag-and-drop checker on the page is free up to a limit, then asks you to sign up free for more free detection credits every month; no credit card required.
- Paid
- price not checked
- Best at
- Flagging all 15 real photos as real in NewsGuard's test.
- The catch
- The same test found it identified only 5 of 15 heavily altered images (33 percent). Its founder said the base model is made for fully AI-generated or heavily edited images.
- web
- api
Upload an image to Gemini and ask "Is this AI-generated?"; Gemini checks for the SynthID watermark.
- Free
- Google says it is available to everyone in the Gemini app; plan limits not checked.
- Paid
- price not checked
- Best at
- Checking images made or edited with Google AI for its SynthID watermark.
- The catch
- Per Google's post it verifies Google AI images only; checking content from other makers depends on future C2PA support.
- Gemini app
Drag a file in to inspect its Content Credentials, the signed record of who made it and how it changed.
- Free
- No price or account is mentioned on the Verify page.
- Paid
- price not checked
- Best at
- Showing a tamper-evident creation and editing history when the file carries one.
- The catch
- The page itself says Content Credentials are still rolling out, so an image may have nothing to view. No credentials does not mean an image is real or fake.
- web
How they compare
Which one suits you, and why.
If you want no login
Hive AI Detector is the easy one here, because the Chrome extension is free with no login. You can check an image by right-click, upload or paste. It also tries to predict which generator made the image, and it covers video too.
What you give up is certainty. It shows a likelihood only. In NewsGuard's test Hive flagged none of 15 real photos, which is good, but it caught only 9 of 15 heavily manipulated ones. So the free, quick option missed a fair few edited images. A Hive result was also cited online to wrongly call a real video fake, which is a reminder not to lean on these scores.
If you want to know which generator made it
AI or Not is the pick here. You upload an image and get a generator breakdown, so you're not just told "AI" but which tool probably made it. It caught all 15 heavily manipulated images in NewsGuard's test, where Hive caught 9.
The cost is false alarms. In that same test it flagged 1 of 15 real photos as AI, and its CEO said low image quality can sometimes cause a false positive. It's mainly sold as an API, with deepfake checks alongside the image checker.
If you want to avoid false alarms on real photos
Sightengine is the one for this. It's a drag-and-drop checker that reports how likely an image is AI-generated and whether faces were manipulated, and it flagged all 15 real photos as real in the test. If your main worry is wrongly accusing a genuine picture, that matters.
But it identified only 5 of 15 heavily altered images (33 percent). Its founder said the base model is made for fully AI-generated or heavily edited images. The free checker has a limit, after which it asks you to sign up for more free monthly credits; we couldn't confirm a price.
If the image came from Google AI
Upload the image to the Gemini app and ask "Is this AI-generated?" Gemini checks for the SynthID watermark, which covers images made or edited with Google AI. Google says it's available to everyone in the app.
The catch is simple: per Google's post it verifies Google AI images only. Checking content from other makers depends on future C2PA support, so it can't vouch for an image from anywhere else.
If you want an image's editing history
Content Credentials Verify lets you drag a file in and inspect the signed record of who made it and how it changed. When a file carries one, that's a tamper-evident history, which is a different kind of evidence from a percentage guess.
The page itself says Content Credentials are still rolling out, so plenty of images will have nothing to view. No credentials doesn't mean an image is real or fake. It just means there's nothing to read.
If you're also wondering about text, video or voice, that's a different question again, because each of those needs its own kind of checker. I've covered it in an app that can detect AI, including which tools handle several types at once. None of them give proof, though, so it's worth knowing what to expect before you trust one.
When paying is worth it
Hive's extension is free with no login, and we found no consumer price for Hive. AI or Not starts you off with $5 in free credits, including 20 AI image checks. After that the Pro plan is $5 per month and includes $10 of credits, with image checks at $0.02 each on the pricing page.
So paying makes sense if you're checking images regularly. For the occasional one-off, the free credits cover 20 image checks before you'd need to pay.
What to watch for
The catches worth knowing first.
NewsGuard's May 2026 audit found five tools collectively calling real photos AI 13.33 percent of the time, and one tool 40 percent. So treat a result as a lead, not proof.
AI or Not's 98.9% accuracy figure is the vendor's own, and its CEO said low image quality can sometimes cause a false positive. A vendor's own figure is a claim, not an independent test.
And remember the Content Credentials point: an empty result there says nothing either way. Also, the audit used 45 images (15 real photos and two edited versions of each), so its figures are one small test, not a ranking.
The same shortlist, as a table
| Thing | Free tier | Paid from | Platforms | The catch |
|---|---|---|---|---|
| Hive AI Detector | The Chrome extension is free with no login (Chrome Web Store listing). | not checked | chrome, web, api | It shows a likelihood only. In NewsGuard's test Hive flagged none of 15 real photos but caught 9 of 15 heavily manipulated ones, and a Hive result was cited online to wrongly call a real video fake. |
| AI or Not | $5 in free credits to start, including 20 AI image checks. | $5 per month (Pro, includes $10 of credits) | web, api | In that same test it flagged 1 of 15 real photos as AI; its CEO said low image quality can sometimes cause a false positive. Its 98.9% accuracy figure is the vendor's own. |
| Sightengine AI image detector | The drag-and-drop checker on the page is free up to a limit, then asks you to sign up free for more free detection credits every month; no credit card required. | not checked | web, api | The same test found it identified only 5 of 15 heavily altered images (33 percent). Its founder said the base model is made for fully AI-generated or heavily edited images. |
| Google SynthID in the Gemini app | Google says it is available to everyone in the Gemini app; plan limits not checked. | not checked | Gemini app | Per Google's post it verifies Google AI images only; checking content from other makers depends on future C2PA support. |
| Content Credentials Verify | No price or account is mentioned on the Verify page. | not checked | web | The page itself says Content Credentials are still rolling out, so an image may have nothing to view. No credentials does not mean an image is real or fake. |
How we checked
What we read, and what we haven't tested yet.
On 2 October 2026 we read the vendor pages listed under sources: the Hive extension's Chrome Web Store listing and Hive's detection page, AI or Not's home and pricing pages, Sightengine's detector page, Google's Gemini announcement, the Content Credentials Verify page and the Content Authenticity Initiative FAQ, plus NewsGuard's published audit. We did not hands-on test any detector.
If none of these fit
The manual way.
Use more than one check. Look for Content Credentials or a SynthID answer, run a second detector, and look for the image's original source. The free checks above can disagree. NewsGuard's audit used 45 images (15 real photos and two edited versions of each), so treat its figures as one small test, not a ranking.