Skip to content

Text · Image · Audio · Video

Analyze AI‑Generated Content

Analyze text, images, audio and video using multiple layers of technical, algorithmic and AI-powered analysis.

0 / 50 000

Up to 50 000 characters · Statistical analysis from 150 words

Analysis levels

Text and images are processed in memory on our server and deleted immediately after analysis. Audio and video files are analyzed directly in your browser and never leave your device. Nothing is shared with third parties.

  • Multi-layer facts, indicators, provenance
  • No storage files deleted after analysis
  • On-device audio & video stay in your browser
  • Explained every signal shown with its value

Methodology

Three layers of analysis

Each layer is independent and clearly separated in your report, so you always know what is a fact, what is an indicator and what is an estimate.

Free

Level 1 — Technical analysis

Facts read directly from the file: format, codecs, resolution, duration, bitrate, EXIF and XMP metadata, color profile, encoding and text statistics.

Free

Level 2 — Algorithmic analysis

Statistical and signal-processing measurements computed without any AI model: sentence rhythm, lexical diversity, error level analysis, spectrum bandwidth, frame changes.

Coming soon

Level 3 — AI-powered analysis

Specialized detection models from named providers, with credits. Each provider and its data policy will be shown before any content is sent.

What we check

Built for every kind of content

Text

  • Sentence length variation
  • Vocabulary diversity (MATTR)
  • Repeated phrases and openings
  • Assistant-style phrases
  • Hidden Unicode characters

Images

  • EXIF, XMP and ICC profile
  • C2PA Content Credentials
  • Generator parameters in metadata
  • Error level analysis map
  • JPEG quality and subsampling

Audio

  • Codec, bitrate, sample rate
  • Effective bandwidth
  • Digital silence detection
  • Pause regularity
  • Noise floor and dynamics

Video

  • Codec, FPS, keyframes, bitrate
  • Device and encoder metadata
  • C2PA provenance
  • Frame sampling and changes
  • Audio track analysis

Transparent reports

Facts, indicators and a cautious conclusion

We never turn an estimate into a certainty. Every report separates what was observed from what was inferred, and explains why.

Read the methodology
  1. FactsInformation actually present in the file.
  2. IndicatorsMeasured signals, each rated none, weak, moderate or strong.
  3. AI estimationProbabilistic model output, when Level 3 is available.
  4. ConclusionA cautious interpretation with its limitations.

Blog

Latest articles

All articles

FAQ

Frequently asked questions

What can TheiBusiness analyze?

TheiBusiness AI Analyzer examines four kinds of content: text (pasted or from a public web page), images, audio and video.

For each one, it reports the technical facts it can observe in the file (Level 1) and a set of algorithmic indicators computed from the content itself (Level 2). AI-powered analysis through specialized providers (Level 3) is coming soon.

Current limits

Text up to 50 000 characters
Images JPG, JPEG, PNG, WEBP, GIF — up to 20 MB
Audio MP3, WAV, M4A, AAC, FLAC, OGG, OPUS — up to 200 MB
Video MP4, MOV, M4V, WEBM — up to 200 MB
Can the analyzer prove that content was made by AI?

No. No detector, including ours, can prove that content was created by a human or by AI. The analyzer reports indicators: observations that are more or less common in generated content.

Results use cautious wording such as No significant indicators detected, Some unusual indicators detected, Multiple indicators detected or Analysis inconclusive. We never display “100% AI”.

The only exception is provenance metadata: when a file declares itself as AI-generated (for example through C2PA Content Credentials or IPTC metadata), the report says so, and notes that the declaration was not cryptographically verified.

Never use a result on its own to accuse someone of using AI.

Are my files stored?
  • Text and images are sent over HTTPS to our server, analyzed in memory and deleted as soon as the analysis finishes. We do not keep a copy.
  • Audio and video are analyzed directly in your browser. The file never leaves your device.
  • URLs: the page or image is downloaded to a temporary file, analyzed and deleted immediately. Any leftover temporary file is removed automatically.

Nothing is sent to a third-party AI service. If AI-powered providers are added in the future, each one will be clearly identified before you send content to it.

Which file formats and sizes are supported?

These limits are read live from our configuration:

Text up to 50 000 characters
Images JPG, JPEG, PNG, WEBP, GIF — up to 20 MB
Audio MP3, WAV, M4A, AAC, FLAC, OGG, OPUS — up to 200 MB
Video MP4, MOV, M4V, WEBM — up to 200 MB

Audio and video decoding happens in your browser, so a format your browser cannot decode (for example some MOV codecs in Firefox) may only receive Level 1 container analysis.

See all questions

Check your next file in seconds

Technical and algorithmic analysis are free. No account required.