AI Text Similarity Detector — free text similarity & ChatGPT paraphrase checker
Paste an original text on the left and a suspected AI-rephrased version on the right. A five-layer pipeline scores semantic similarity, concept preservation, lexical overlap, AI fingerprints and sentence-level matches — then classifies the result on a 6-tier scale from IDENTICAL through AI CONFIRMED to ORIGINAL CONTENT. No upload, no signup, runs in your browser.
Running 5-layer AI rephrasing analysis…
—
Run an analysis to see the verdict.
When to run an AI rephrasing check
Compare two texts whenever you need to know whether the second is an AI-paraphrased version of the first — works for essays, articles, social posts, AI-rewritten drafts and bulk-spun content.
Spot AI-rewritten essays
Catch student work that was passed through ChatGPT, Claude or a paraphraser to disguise an original source.
Audit news & blog drafts
Compare a draft against a source article to confirm it wasn't just AI-paraphrased from a published piece.
Test paraphraser output
Measure how heavily QuillBot, Wordtune, or "humanizers" rewrite your text and how much of the meaning survives.
Detect scraped content
Identify AI-rewritten content scrapers that take your article and lightly rephrase it before republishing.
Verify your own paraphrase
Check that your paraphrase is genuinely different from the source — not an accidental near-copy.
Build a content claim
Document semantic and AI-pattern evidence when you need to file an attribution or copyright complaint.
How to detect AI rephrasing (4 steps)
Paste the original text
Drop your source article, draft or human-written reference into the left editor.
Paste the suspect text
Paste the text you suspect was AI-rephrased, paraphrased or rewritten into the right editor.
Analyze
Click Analyze. The detector runs six stages — preprocessing, feature extraction, semantic analysis, AI pattern detection, sentence mapping and final assessment — in a few seconds.
Review & export
Read the verdict band, six headline metrics, sentence-level matches, AI fingerprints and the detailed-metrics breakdown. Export a TXT or HTML/PDF report when you need evidence.
Frequently asked questions
An AI rephrasing detector compares two texts to determine whether one is an AI-paraphrased version of the other. HCODX's detector uses a 5-layer pipeline: semantic embeddings from the Universal Sentence Encoder, concept preservation, lexical overlap, AI fingerprints (formal phrasing, transition word frequency, uniform sentence lengths), and sentence-by-sentence mapping with length tolerance.
It looks for a specific signature — high semantic similarity (the texts mean the same thing) combined with low lexical similarity (the words are different) plus AI fingerprints. When that gap is large and the same concepts are preserved, the score is boosted toward an AI verdict. When both texts simply discuss the same topic without sharing structure, the score is reduced.
Yes. The HCODX AI Text Similarity Detector is 100% free, ad-free, and requires no signup or login. There are no daily limits or paywalls.
No. Both texts stay in your browser. Embeddings and similarity computations run locally with TensorFlow.js and the Universal Sentence Encoder. The model is downloaded once and cached, after which the tool works offline.
IDENTICAL (≥96%) means the texts are essentially the same. AI CONFIRMED (≥85% with a wide semantic-lexical gap) means strong evidence of AI rephrasing. AI LIKELY (≥75%) and AI POSSIBLE (≥65%) indicate progressively softer evidence. SEMANTIC SIMILAR (≥39%) means the texts overlap in meaning without strong rephrasing signals. ORIGINAL CONTENT (<39%) means no significant rephrasing detected.
Yes — detection is model-agnostic. It targets the rephrasing signature itself (high semantic + low lexical + AI fingerprints) rather than any particular model. It works on text rewritten by ChatGPT, Claude, Gemini, Llama, DeepSeek, Mistral and most paraphrasing tools.
The tool detects identical input and short-circuits to a 100% IDENTICAL verdict — it doesn't inflate the AI score when the text hasn't actually been rewritten.
Yes. After each run you can export a plain-text report or a printable HTML/PDF report listing the verdict, all six metrics, sentence mappings, AI patterns and an executive summary.
About this AI rephrasing detector
HCODX's AI Text Similarity Detector is a free, ad-free, no-signup tool that compares two texts to detect AI paraphrasing, rewriting and semantic similarity directly in your browser. It runs on top of TensorFlow.js and the Universal Sentence Encoder — nothing is uploaded.
The 5-layer pipeline
- Layer 1 — Semantic similarity. Sentence embeddings on cleaned text capture meaning regardless of wording.
- Layer 2 — Concept preservation. Jaccard overlap on extracted key terms tells us whether the same concepts survived the rewrite.
- Layer 3 — AI fingerprints. Counts of formal/transitional phrasing, sentence-length uniformity, structural markers — patterns characteristic of LLM output.
- Layer 4 — Sentence mapping. Each suspect sentence is matched to its best original counterpart (length-filtered, 25% tolerance), then flagged as
AI rephrasedwhen semantic ≥0.5 and lexical ≤0.4. - Layer 5 — Topic vs. derivation. If both texts discuss the same topic but show no real derivation signal, the score is reduced — coincidence isn't plagiarism.
Score formula
The final score is semantic-first: 0.7 · semantic + 0.3 · (cosine + jaccard + edit)/3, with a +0.1 bonus when the semantic-lexical gap exceeds 0.6 (and +0.05 when it exceeds 0.5). Identical inputs (similarity ≥0.999) shortcut to 100%.
Verdict bands
≥ 96%— IDENTICAL. Essentially the same text.≥ 85%(gap > 0.6) — AI CONFIRMED. Strong rephrasing signature.≥ 75%(gap > 0.55 or pattern signals) — AI LIKELY.≥ 65%(gap > 0.51) — AI POSSIBLE.≥ 39%— SEMANTIC SIMILAR. Topic overlap without strong derivation evidence.< 39%— ORIGINAL CONTENT.