Credit for Creators,
Control for AI
- Verified sources for images, videos and sounds
- Verified creator details
- Flagged AI usage
- Freedom to think again
Flags
Each flag says what was found. Hover over a flag, or tap it, to see why it's there.
- Human-madeVerified as captured on a camera or phone, with no AI.
This verifies the photo, not what’s in it: a photo of a painting or a screen doesn’t verify the painting or what’s on the screen.
Supported by these camera models. - Likely Human-madeDetails suggest a person made this, such as a camera and its settings, or illustrative tools and the artist’s name, but we couldn’t verify the data. Illustrations and digital art can’t be verified higher than Likely Human-made, because there are no public verification tools for them yet.
- InformationNot enough information to verify the content, but we will share what we found.
- AI AssistedPartly made with AI, edited with AI (including an AI assistant), labelled as AI by the website, or reported by you.
- AI GeneratedMade by AI, as marked by the AI itself.
This includes some content edited on AI platforms, as the AI recreates the pixels of the original.
Blurring
| Setting | Blurs |
|---|---|
| Relaxed | AI Generated |
| Balanced (to start with) | AI Generated and AI Assisted |
| Human only | Everything except Human-made and Likely Human-made |
The flags never change with this setting. Press Ctrl+U (⌘U on a Mac) to see everything on a page, and again to blur it.
Less AI where you don't want it.
A pause before AI sites
Opening ChatGPT, Gemini, Claude, Midjourney and over 150 other AI sites shows a short pause screen first.
It says what the site is and why it was paused.
Then you choose: go back, carry on, pause it for two hours, or tell it this isn't an AI site.
Less AI in Google
Web results only, without AI Overviews or AI Mode.
Chatbots and bot accounts
Chat windows that answer with AI get a tag, and so do AI accounts on social sites, like Grok's replies on X.
AI in text
When a website says its own text was written with AI, or a page is a shared chatbot conversation, ThinkAgain tells you.
Deep Check
Right-click any picture for a full report: its details, invisible watermarks, AI icons and who made it, with each AI company's own check, the picture copied for you to paste.
AI on YouTube
Videos that YouTube labels as altered or synthetic, or whose titles say they were made with AI, are flagged and paused.
Right-click a channel's name to hide all its videos.
Checks run on your computer.
No account, no ads and no tracking. ThinkAgain never collects your browsing history.
Exceptions
These checks send data about an image to other companies.
None of it includes your name, email address or browsing history.
| Who receives it | Data sent | Image sent |
|---|---|---|
| Adobe | Off by default | Off by default |
| Digimarc | Never | Off by default |
| Watermark registry | Off by default | Never |
| Google, Meta or OpenAI | Never | If you paste it |
The privacy policy lists everything ThinkAgain sends.
Questions
Why are some pictures missing flags?
Most pictures online carry no Content Credentials or AI details, and many websites remove them when pictures are uploaded.
With nothing to check, ThinkAgain leaves the picture alone rather than guessing.
Why isn't this AI picture flagged?
ThinkAgain only flags what it can show evidence for.
An AI picture with that evidence removed looks like any other picture.
Report it, and it's hidden for you straight away.
What does ‘From the file’s details’ mean?
The details were found in the file, but anyone could have written or changed them, such as a camera name or an editing history.
ThinkAgain shows them as information, not proof.
Only Content Credentials that check out count as verified.
Will it slow down my browser?
Pages load as normal.
ThinkAgain checks pictures as they appear, so a flag can take a moment to show.
Which browsers are supported?
Google Chrome on a computer.
Phones and tablets can't run Chrome extensions.
Does ThinkAgain work with the EU AI Act and California’s AI laws?
ThinkAgain reads the AI marks these laws ask AI companies and platforms to add.
It can't see marks that were never added or have been removed.
More answers on the support page.