Get in Touch

Free · No sign-up · File stays on device

Free AI Image & PDF Checker

A free AI image checker and free AI PDF checker for provenance: drop a JPEG, PNG, WebP or PDF and read its Content Credentials, who signed it, what made it, its recorded edit history and any declared AI origin, entirely in this browser. It reads evidence, not pixels.

Runs in your browser

One file. An evidence sheet, not a verdict.

Check an image or PDF with the free AI image checker

Drop a JPEG, PNG, WebP or PDF here or

Its Content Credentials (C2PA signed origin and edit history) are read with the official Content Authenticity Initiative engine, entirely in this browser — the engine's remote fetching is switched off, so the file never leaves this device. 20 MB maximum.

The evidence sheet

Drop a file above and this sheet fills in with what its signed credentials actually say: who signed it, what made it, when, how many recorded edits and ingredients, and whether the signature validates locally. A file with no credentials is reported as exactly that — evidence, not a judgement.

Read it correctly

What a free AI image detector like this can, and cannot, tell you

No evidence found does not mean real. Most files carry no Content Credentials, and credentials that were present are easily lost — screenshots, re-encodes and most platforms strip them. An empty sheet is evidence of nothing either way.

It can tell you

Whether the file carries C2PA Content Credentials; who signed them; what software or model recorded itself as making or editing the file; when it was signed; how many assertions and ingredients the manifest carries; and whether the signature validates locally.

It cannot tell you

Whether the signer is trustworthy — certificate trust lists are not checked in this browser release, and the sheet prints that beside the validation state. It is not a pixel-based AI media detector: it gives no AI-image likelihood, no metadata forensics and never a percentage, because a number that collapses on the most common generators would mislead in the direction of looking authoritative.

What comes next

In roadmap order: trust-list verification, so a valid signature can also read trusted signer or unknown signer; provider verification where a checkable answer exists for images or audio; and a clearly probabilistic AI-image likelihood channel, published with its measured error rates.

Provider comparison

Compare published text-watermark support

See, source by source, what OpenAI, Google and Anthropic say about marking text and whether they offer a public text check.

Every cell verified against the provider's own page on 29 August 2026. Re-date on republish. If this date is older than the page carrying it, treat the panel as unverified.

Marks its own text?What the provider has published
Can anyone outside verify it?Whether a public check exists for text
Google Gemini Gemini app and web
Published as marked

Google says Gemini's text output carries SynthID-Text, and has done since 14 May 2024.

deepmind.google/science/synthid/ai-generated-text/
Verified 29 August 2026 · review monthly

No public text verifier

Google's public checking covers images, video and audio. Text is not an accepted input.

deepmind.google/models/synthid/
Verified 29 August 2026 · review monthly

Anthropic Claude All shipping models
Committed, not demonstrated

Anthropic says Claude models launched on or after 2 August 2026 support marking at launch, and that it is working to add marking to models released earlier.

No Claude model has launched on or after that date. Opus 5 launched 24 July 2026, Sonnet 5 on 30 June. Anthropic publishes no per-model status.

support.claude.com … how-claude-marks-ai-generated-content
Verified 29 August 2026 · review monthly

Announced, not released

Anthropic says it will soon offer a watermark detection API. It has not shipped one.

anthropic.com/news/claude-text-watermark
Verified 29 August 2026 · review monthly

About 2 August 2026. That is the date the EU AI Act's Article 50 transparency obligations began to apply. Anthropic inherited it as a legal boundary rather than choosing it as a rollout date. It is not a date on which Claude output began carrying a mark, and no such date has been published. digital-strategy.ec.europa.eu … guidelines-ai-transparency-obligations · Verified 29 August 2026

OpenAI ChatGPT Text output
Not publicly documented

OpenAI's provenance documentation covers images and audio only; no text watermark is publicly documented.

developers.openai.com/api/docs/guides/content-provenance
Verified 29 August 2026 · review monthly

No public text verifier

Its provenance check endpoint accepts image and audio files. Text is not an accepted input.

developers.openai.com/api/docs/guides/content-provenance
Verified 29 August 2026 · review monthly

Media provenance questions

Questions about the image and PDF evidence sheet

What does this free AI image checker actually check?

One channel, answered honestly: C2PA Content Credentials. A JPEG, PNG, WebP or PDF is read with the official Content Authenticity Initiative engine, entirely in your browser, and the sheet reports what the credentials say — who signed the file, what software or model made or edited it, when it was signed, how many recorded assertions and ingredients it carries, and whether the signature validates locally. It reads evidence — signed credentials and declared metadata — not pixels, and it never prints an AI percentage.

Does my file leave this device?

No. This file-checking route runs in your browser. Remote manifest and certificate-status fetching is switched off, so no request carries any part of the selected file. The file stays on this device and is not stored or sent to analytics.

The sheet found no Content Credentials — does that mean the file is AI, or real?

Neither. No evidence found does not mean real, and it does not mean AI-generated either. Most files on the web carry no credentials, and credentials that were present are easily lost: screenshots, re-encodes and most platforms strip them. Absence is recorded as evidence without a judgement, which is exactly what it is.

The signature validated — does that prove the file is authentic?

It proves less than it sounds like. Local validation checks that the manifest is intact and consistently signed. Certificate trust lists are not checked in this browser release, so this page cannot say whether the signer is who they claim to be or whether they are on the C2PA trust list — and it says so on the sheet rather than in small print.

Why is there no “N% AI” score?

Because a single percentage is not defensible. Public AI-image detectors perform worst on exactly the current generators people actually use, and a number that collapses on the most common case misleads in the direction of looking authoritative. If a likelihood channel ships here it will be labelled probabilistic, shown beside its measured error rates, and never presented as a verdict.

Does this free AI image detector analyse pixels?

No. It reads signed C2PA Content Credentials and declared metadata, not pixels, so it is not a pixel-based AI media detector and never prints an AI percentage: a guess that collapses on the most common generators would only look authoritative.

Is there a free AI PDF checker here?

Yes. Drop a PDF and the free AI PDF checker reads its Content Credentials and declared metadata in your browser, reports nothing declared as exactly that, and never scores the text: to check the writing itself, paste the text into the free AI content checker.

What is planned next for this page?

Three channels, in roadmap order: certificate trust-list verification, so a valid signature can also say trusted signer or unknown signer; provider verification where a provider offers a checkable answer for images or audio; and a clearly probabilistic AI-image likelihood reading. Each will state what it measured and what it cannot say, like everything else in this suite.

Free · Runs in your browser · File stays on device

Read what a file's signed credentials actually say.

Drop a JPEG, PNG, WebP or PDF above: the official Content Authenticity Initiative engine reads it on your own device, and the sheet reports evidence, never a verdict. For a free AI PDF detector reading of a document's text, paste the text into the free AI content checker.