AI content verification

Detect AI-generated content. See the evidence.

Check text, images, audio, video, and code for signs of AI generation. Get a first result before creating an account.

Text Image Audio Video Code

Your first result, no sign-up needed

Five content types

Check what you read, see, and hear

Try it before signing up

Your first result opens right here

Keep checking in your account

Save results and return to previous scans

Inside a result

See what is behind the score.

Start with the probability, inspect the supporting signals, then put them in context.

1.
Read the overall signal
See whether the sample has stronger AI-associated or human-origin signals.
2.
Look at the details
Use the highlighted regions and timestamps to find what deserves a closer look.
3.
Check the source
Consider where the content came from and how it was edited before drawing a conclusion.

A reason to investigate, not a verdict on who made it.

Result summary

Sample analysis

AI-associated signals are stronger

Estimated AI probability

AI 84%Human 16%

voice-note.wav00:18.4

00:0000:09.200:18.4

“Send the document this evening.”

Example annotation: compare the even spacing between words with nearby speech.

Interface example. Scores, waveform, and notes are illustrative, not a live analysis of a recording.

Five ways to check

A closer look at every format.

The same workflow, with details that make sense for the content you are checking.

TXT

Text

Look at the writing in context, with highlighted passages and an explanation of what to review.

Open this detector

Interface examples, not live scan results.

article-draft.txt

The old library is being redesigned around the people who use it. The plan adds quiet desks and an open reading room.

This approach improves access. This approach also reduces waiting times.

The first changes are planned for autumn, starting with the entrance and the ground floor.

Highlighted passage · Repeated openings

Both sentences begin the same way. This is a pattern to review in context, not proof of AI authorship.

When to use ZeroTrue

Check before you trust, share, or act.

A photo in your feed. An unexpected voice message. A draft that does not sound quite right.

Images and video

Before you share what you see

Take a closer look at an image before passing it on. Check its source alongside the result.

AI-generated image of a concrete house beside a still lake, with its window reflected in the water

Voice and audio

When a voice message feels wrong

Check the audio, then verify with the sender through a channel you already trust.

“Send the document this evening.”

Text and code

When words need a second look

Look at passages in context. Repeated patterns alone do not establish who wrote them.

“This approach improves access. This approach also reduces waiting times.”

Apps and integrations

Check content where you are.

On your phone, in Telegram, with an AI assistant, or directly from your code. Choose what fits your day.

On your phone.

Check content in the ZeroTrue app for Android and iOS.

Mobile apps

AndroidiOS

Right in Telegram.

Send content to the bot or open the Mini App to use the checking interface inside Telegram.

Two ways in

BotMini App

Illustrated interfaces with example results, not app screenshots.

MCP

Connect content checking to an AI client that supports MCP.

Connect MCP

Agent Skill

Give your AI agent a ZeroTrue skill in the tools you already use.

Install the skill

SDK

Build checking into your application or automation in your language.

Python TypeScript Go

Your first check is still available right here on the website.

API documentation

Know the limits

Useful signals. Honest limits.

Detection can guide a closer review. It cannot establish authorship or intent on its own.

How to read a ZeroTrue result

It can help you

Prioritize content that needs closer review
Compare AI-associated and human-origin signals
Locate regions or moments worth inspecting

It cannot prove

Who created a piece of content
The intent behind its creation or use
That one score is correct in every context

Sample quality, compression, editing, language, and generator shifts can all affect detection output.

Methods and model signals

How detection works and what its signals mean.

Benchmarks and evaluation

How to read evaluations, thresholds, and false positives.

Security and data handling

What to review before uploading sensitive content.

ZeroTrueSample workspace

Workspace

Overview

Scan history

Reports

Recent scans

New scan

AI probability

lake-house.jpg

Image · Just now

78%

voice-note.wav

Audio · 12 min ago

84%

article-draft.txt

Text · Yesterday

61%

Full reports are available alongside each saved scan.

Your next check

Keep your results in one place.

Open full reports, return to earlier checks, and start the next scan from your ZeroTrue account.

Carry the current result into your account
Open detailed evidence and signal context
Return to saved scans and reports

Create a free account

Already checked something? Continue from your result above to keep that scan.

For developers

Integrate in minutes, not weeks.

Use one request pattern for text, images, audio, video, and code. Review the public quick start, then create a key when you are ready to integrate.

Free API access to start
Webhooks for async processing
SDKs for Python, TypeScript, and Go

Read the Documentation

POST/api/v1/check

curl -X POST https://app.zerotrue.app/api/v1/check \
  -H "Authorization: Bearer {{API_KEY}}" \
  -H "Content-Type: application/json" \
  -d '{
    "input": {
      "type": "text",
      "value": "Check this sample for patterns associated with AI-generated writing."
    },
    "is_private_scan": true
  }'

Example result fieldsJSON

{
  "ai_probability": 0.73,
  "human_probability": 0.27,
  "result_type": "ai"
}

Free to start

Start with one check.

Try the browser scanner first. Create a free account when you want to save results and keep checking.

Start scanning free See limits and details

Free account

For everyday checks, saved reports, and exploring the API.

100 detections per month
Web dashboard
API access

Before you start

A few useful answers.

What you can check, what a result means, and what happens to your content.

How accurate is AI content detection?

Accuracy changes with the content type, generator, language, sample quality, editing, compression, and decision threshold. Review modality-specific evaluation data and false-positive rates rather than relying on one universal percentage.

Can ZeroTrue prove that content is AI-generated?

No detector can establish authorship or intent from a score alone. ZeroTrue provides probabilistic evidence and supporting signals that should be combined with source context and human review.

What content can I check?

ZeroTrue supports text, common image formats, speech audio, video, and source-code files. The scanner shows the currently accepted formats and size limits before upload.

What happens to uploaded content?

Data handling can depend on the selected product mode and plan. Review the current privacy and security terms before submitting sensitive content or using ZeroTrue in a regulated workflow.

Can edited or compressed content still be checked?

Yes, but editing, screenshots, transcoding, noise, and social-media compression may weaken or change detectable signals. These transformations should be considered when interpreting the result.

Can I try ZeroTrue for free?

You can run a first browser check before creating an account. The current free plan includes a monthly detection allowance for continued use.

Is there an API for developers?

Yes. ZeroTrue provides one API pattern for text, images, audio, video, and code, with public request examples and a free starting allowance.

Something does not look right?

Paste the text or upload a file. Your first result is a good place to start.

Start a scan Explore the API