How does the AI Detector work?

AADDYY's AI Detector uses a dual-signal approach: the Raidar method (measuring how much an AI rewrites the text — AI barely edits AI-generated text) combined with a direct classifier that identifies specific AI patterns. Both signals are combined for a more accurate result than either approach alone.

What do the 4 classifications mean?

Human ( 80%): almost entirely produced by AI.

What signals does it detect?

The detector looks for: uniform sentence lengths, formulaic transitions ('Furthermore', 'It is worth noting'), low rhythm variation, predictable word choices, overly linear structure, and lack of personal voice. Each flagged sentence shows which pattern triggered the flag.

What is ESL Fairness Mode?

Non-native English speakers often write in patterns that resemble AI — simple vocabulary, uniform sentence lengths, formulaic transitions — because these are taught in ESL education. ESL Fairness Mode raises the AI threshold by 15 points to reduce false positives for non-native writers.

What's the minimum text length?

At least 50 words (roughly 300-400 characters) for reliable results. Very short texts don't have enough signal for accurate detection. The tool will warn you if your text is too short.

What should I do if my text is flagged as AI?

Click 'Humanize →' to send the text directly to AADDYY's AI Humanizer. This rewrites the AI-sounding sentences to sound naturally human. After humanizing, you can run the detector again to confirm the score improved.

AI Detector

AADDYY's AI Detector identifies AI-generated content using dual-signal analysis — showing exactly which sentences triggered detection and why.

Detection Result

Paste text to analyze.

How it Works

Paste your text

Copy and paste any text you want to check — minimum 50 words for reliable results. Supports up to 10,000 characters.

Enable ESL mode if needed

If the author is a non-native English speaker, toggle ESL Fairness Mode to adjust thresholds and reduce false positives.

Review the analysis

See the 4-tier classification, which sentences triggered detection, what signals were found, and best-guess model attribution.

Use Cases

Content publishing

Verify that articles, blog posts, and content submissions are written by humans before publishing to maintain editorial standards.

Academic integrity

Check student submissions for AI-generated content as part of your review process alongside your academic judgment.

Freelance hiring

Verify that writing samples from freelancers or job applicants are genuinely their own work.

Personal checking

Check your own AI-assisted writing to understand how it scores before submitting to publications or platforms with AI policies.

Also in this category

Questions.

Still have questions? We're here.

Contact Support