What is a good Turnitin similarity score?

Most universities consider a similarity score under 20-25% acceptable. However, this varies by institution and assignment type. A 15% similarity score with properly cited quotes is perfectly normal, while a 15% score from uncited paraphrasing could be problematic.

What Turnitin AI score triggers a review?

Turnitin flags submissions with an AI score of 20% or higher. However, Turnitin recommends that instructors investigate any score above 0% rather than relying on a fixed threshold, as the tool is meant to inform human judgment, not replace it.

Can you have a high similarity score and low AI score?

Yes, absolutely. A student who copies and pastes from published sources (high similarity) while writing their own original text for the rest (low AI) would see this pattern. The two scores measure completely different things.

Can you have a low similarity score and high AI score?

Yes. AI-generated text is original (not copied from existing sources), so it typically has a very low similarity score. But it will score high on AI detection. This is actually the most common pattern for AI-generated submissions.

Does Turnitin show professors the AI score automatically?

Yes. Since 2024, Turnitin's AI detection is enabled by default for most institutional accounts. Professors see both the similarity score and the AI score in the Turnitin report unless their institution has specifically disabled AI detection.

Educational March 8, 2026 12 min read

Turnitin Similarity Score vs AI Score: What's the Difference?

Students see two scores on every Turnitin report. One measures plagiarism. The other measures AI. Understanding the difference could save your grade.

Reviewed by Dr. Sarah Chen · Founder & CEO

Key Takeaways

Turnitin's similarity score measures text matching against published sources (plagiarism). The AI score measures statistical patterns associated with AI generation. They are calculated independently.
A similarity score under 20-25% is generally acceptable. AI scores above 20% typically trigger an instructor review.
You can have a high similarity score with a low AI score (copied from sources) or a low similarity score with a high AI score (AI-generated original text).
Professors see both scores by default since 2024. The AI score includes a sentence-by-sentence highlight showing which parts were flagged.
To reduce your AI score without affecting content quality, use an AI humanizer to address the statistical patterns detectors measure.

The Two Scores Explained

When your professor opens your Turnitin report, they see two separate scores. Many students (and some professors) conflate these two metrics, but they measure fundamentally different things.

Similarity Score

What it measures: How much of your text matches existing sources in Turnitin's database (internet, published papers, other student submissions).

How it works: Text-matching against a database of 1.5+ billion web pages, 100+ million published works, and 300+ million student papers.

What it catches: Copy-paste plagiarism, inadequate paraphrasing, missing citations, recycled student work.

AI Score

What it measures: The probability that your text was generated by an AI language model (ChatGPT, Claude, Gemini, etc.).

How it works: Analyzes statistical patterns like perplexity and burstiness to determine if text follows AI-characteristic patterns.

What it catches: AI-generated text, even if it is completely original and does not match any existing source.

What Professors Actually See

Here is what appears in a professor's Turnitin dashboard when they open your submission:

1. Overall similarity percentage with color-coded highlights showing which passages match which sources
2. AI writing indicator showing a percentage of text flagged as AI-generated
3. Sentence-level AI highlighting where specific sentences are marked as "AI-generated" with varying confidence levels
4. AI writing report with a detailed breakdown of the analysis

Important: since 2024, AI detection is enabled by default for most institutional Turnitin accounts. Your professor sees the AI score automatically unless their university has specifically opted out.

Score Thresholds: What Triggers a Review?

Score Type	Green (Safe)	Yellow (Caution)	Red (Review)
Similarity Score	0-15%	16-25%	25%+
AI Score	0-10%	11-20%	20%+

Note: These thresholds vary by institution. Some universities set stricter or more lenient cutoffs. Always check your institution's specific policy.

Common Score Combinations and What They Mean

Low Similarity + Low AI = Human-Written Original Work

The ideal outcome. Your work is original and written by you. No flags.

High Similarity + Low AI = Plagiarism (Not AI)

You copied from existing sources without proper citation. This is a traditional plagiarism issue, not an AI issue.

Low Similarity + High AI = AI-Generated Content

The text is original (no matches) but shows AI patterns. This is the most common pattern for students using ChatGPT or similar tools.

High Similarity + High AI = AI Content from Trained Sources

Rare, but can happen when AI generates text that closely mirrors its training data or when AI output includes common phrases from published sources.

How to Reduce Each Score

Reducing Your Similarity Score

Use proper quotation marks and citations for all direct quotes
Paraphrase sources in your own words rather than rearranging the original
Include a complete bibliography/works cited page
Exclude bibliography and quoted material in Turnitin settings (if your professor allows)

Reducing Your AI Score

Write your own content from scratch (most reliable method)
If using AI as a drafting tool, substantially rewrite the output in your own voice
Add personal examples, opinions, and specific experiences that AI would not generate
Use an AI humanizer to address statistical detection patterns while preserving content quality

What to Do If You Get Falsely Flagged

Turnitin's AI detection has a documented false positive rate of approximately 4% at the document level. If you wrote your work yourself and received a high AI score, you have the right to appeal the result.

Steps to take: contact your professor immediately, provide your drafts and revision history, explain your writing process, and reference Turnitin's own documentation about false positive rates.