Direct answer
Direct Answer - Turnitin displays AI writing scores below 20% as *% (asterisk) rather than a specific numeric percentage because the detection model has lower confidence at that threshold [1]. This *% benchmark implies the document is less likely to contain substantial AI-generated prose, but it does not guarantee the text is entirely human-written, nor does it immunize a submission from instructor scrutiny [1][2]. The 20% threshold is a confidence-based reporting boundary, not a pass/fail line for academic integrity.
What Does Turnitin Consider a 20% AI Writing Score Threshold and How Is It Displayed?
Turnitin's AI writing detection model evaluates submitted prose and assigns a percentage representing how much of the document the model identifies as likely AI-generated [2]. When the overall indicator falls below 20%, Turnitin intentionally suppresses the exact number and replaces it with an asterisk *% [1]. This design choice reflects a fundamental principle: at lower confidence levels, displaying a precise figure like "3%" or "12%" could create a false sense of precision that undermines the indicator's reliability [2].
The 20% cutoff is not arbitrary — it is statistically tied to the model's confidence calibration. Turnitin reports a false positive rate of less than 1% for documents scoring 20% or higher, meaning predictions above that threshold carry stronger reliability [1]. Below that mark, the model's confidence diminishes, and the asterisk signals this reduced certainty to educators and students alike [2]. Importantly, the report also highlights specific sentences that contributed to the score, giving instructors granular visibility regardless of the overall percentage [2].
Educators are explicitly advised to treat the AI writing indicator as a starting point for discussion, not as a definitive judgment of academic misconduct [1]. The asterisk display is meant to encourage contextual review and professional judgment rather than automated decision-making based on a single number [2].
Can Instructors Still Flag a Paper If the Turnitin AI Score Is Under 20% or Shows an Asterisk?
Yes — and this is one of the most critical nuances of the 20% benchmark. An *% score does not prevent an instructor from flagging a paper for further review [3]. Turnitin's own documentation emphasizes that the AI writing report is a tool designed to "start conversations, not to make final decisions," and instructors retain full discretion to evaluate submissions as they see fit [3]. The asterisk simply means the model's confidence was too low to assign a numeric percentage — it does not mean the document was certified as human-written [3].
Furthermore, the report provides sentence-level highlights that flag specific passages the model identifies as potentially AI-generated, and these highlights are visible regardless of whether the overall score is above or below 20% [3]. An instructor who sees suspicious flagged sentences can manually review those sections, consult with the student, or apply institutional academic integrity policies irrespective of the asterisk display [3]. Turnitin explicitly recommends that educators "use the AI writing indicator as one of several pieces of evidence" in their assessment [3].
This means a sub-20% score offers no guarantee of immunity from scrutiny. Instructors are trained to consider the full context — writing style, assignment parameters, student history, and the flagged highlights — rather than relying solely on the percentage [3]. Understanding this empowers students to engage more thoughtfully with their reports.
How Can Students Accurately Verify Their Turnitin AI Score Before Submitting to Avoid Misinterpretation?
The most effective strategy for students is to check their drafts using the same Turnitin AI detection tools their instructors use, before the final submission deadline [4]. When students preview their AI writing indicator proactively, they can see whether their text lands above or below the 20% threshold and, crucially, which specific sentences or passages the model flags [4]. This pre-submission insight eliminates guesswork and allows students to address any flagged content — whether it is properly attributed, paraphrased, or discussed with their instructor before the final submission [4].
Turnitin's guidance stresses that "understanding what the AI writing indicator can and cannot tell you is essential" for students navigating this landscape [4]. A pre-submit check clarifies whether that *% result is due to genuine human writing patterns or simply the model's lower confidence in detecting sophisticated AI output. Students can then make informed decisions: if they have used AI tools legitimately (e.g., for brainstorming or editing), they can prepare to discuss their process openly with instructors rather than being caught off guard [4].
Ultimately, proactive verification transforms the 20% threshold from a source of confusion into a practical reference point. Students who understand that *% is not a "pass" but a lower-confidence indicator can take ownership of their academic integrity and avoid the misinterpretation that too often leads to unnecessary academic integrity proceedings [1][4].
Understanding the 20% threshold is only half the equation — the real value comes from knowing your actual AI and similarity scores before your instructor sees them. With Turnitin0, you can preview the exact same Turnitin AI and similarity reports that your university's system generates, giving you clarity on where your draft stands before submission day.
FAQ
Does a Turnitin AI score under 20% mean my paper is safe from being flagged?
No. An *% score means Turnitin's model had lower confidence in its prediction, but instructors can still manually review flagged sentences and apply academic integrity policies regardless of the overall percentage [1][3].
What does the asterisk *% actually mean on a Turnitin report?
The asterisk indicates the overall AI writing prediction fell below the 20% reporting threshold, so Turnitin suppresses the exact number because the model is less confident in low-range predictions [1][2]. It does not mean the document is proven to be human-written.
Can I get in trouble for AI use even if my Turnitin score is under 20%?
Yes. The AI writing indicator is only one data point — instructors consider flagged highlights, writing style, student history, and assignment context [3]. A low score does not override professional academic judgment.
Is there a way to check my Turnitin AI score before submitting to my instructor?
Yes. Services like Turnitin0 allow you to upload your draft and receive the same Turnitin AI and similarity reports that instructors see, giving you full visibility into your scores before formal submission [4].
Should I trust a Turnitin AI score under 20% as proof of human authorship?
No. Turnitin explicitly states that a low confidence prediction "does not mean the document is free of AI-generated text" [1]. The 20% threshold is a reporting boundary, not a certification of authorship.
※ Turnitin0.com - Actual Turnitin AI Report Cover, Score, Flag And Similarity Summary