Direct answer
The Pangram AI detector Chrome extension is a browser add-on from Pangram Labs that scores any text you highlight or open in the browser and returns a probability that the text was written by an AI model [1]. It runs inside Chrome rather than on a submitted file, so it is a fast self-screening tool for drafts, not an institutional report. It cannot reproduce the Turnitin AI writing report your professor sees, because that report is generated from the actual file you submit to your course's learning management system [4].
What Is the Pangram AI Detector Chrome Extension and What Does It Actually Do?
The extension is a thin front end for Pangram's detection model. You install it, and when you are reading a document, an email draft, or a page of text, you ask it to score that text; the extension sends the text to Pangram and returns a score indicating how likely it is to be machine-generated [1]. Pangram positions the product around sentence-level highlighting, so instead of a single opaque number you get a visual map of which spans pushed the score up [2]. That granularity is genuinely useful when you are trying to work out whether one paragraph or the whole document is reading as AI.
It is important to understand what the output means. Pangram's own documentation describes the result as a probability produced by a classifier trained on human and machine text, not as a finding of misconduct [2]. A high score means the wording resembles the patterns the model associates with generated text; it does not establish who wrote anything or whether a rule was broken. The same documentation notes that results depend on the model behind the text and can shift when the text is edited, which is why two detectors can disagree about the same paragraph [2].
Practically, the extension is built for speed and convenience. It works on text you can see in the browser, which makes it handy for checking a paragraph before you paste it into your assignment, or for sanity-checking a section you just rewrote [1]. It is not a document pipeline: it does not take a .docx upload, it does not produce a downloadable report, and it does not tell you anything about similarity or plagiarism [1]. Those are different jobs, and conflating them is where most students get confused.
The most common mistake is treating a browser score as a rehearsal for the real thing. A browser extension evaluates text as it appears on screen, stripped of the file's structure; your institution evaluates a specific uploaded file through a specific system [4]. Those two evaluations can land in different places even when the words are identical, so the extension is best used as an early warning rather than a final verdict [2].
Is the Pangram AI Detector Accurate, and How Does It Compare With Turnitin's AI Detection?
Both tools are classifiers, and both express their output as a likelihood rather than a proof, but they are calibrated and documented very differently. Turnitin publishes its methodology and error rates: its AI writing indicator reports an overall percentage of qualifying prose flagged as AI-generated, alongside a sentence-level breakdown, and Turnitin states that for documents containing at least 20% AI writing the overall false-positive rate is below 1% [3]. That published error rate is what makes the indicator defensible in an academic-integrity conversation, and it is the number your institution is actually relying on [3].
Turnitin is also explicit about the limits of its own signal. When the detector's confidence falls below its threshold, Turnitin displays *% instead of an exact percentage, signalling a low-confidence result rather than a precise finding [3]. Turnitin further cautions that the indicator is not definitive proof of misconduct and should be read in context alongside the assignment, the drafting history, and the student's own account [3]. A responsible reading of any detector score, Pangram's included, follows the same logic: a number is a prompt for a conversation, not a conclusion [2][3].
Where the two diverge most is scope and consequence. Pangram's extension scores whatever text you point it at, which is flexible but also unstandardised — the input is not a submitted artefact and the output is not tied to any institutional process [1][2]. Turnitin's indicator is generated inside the Similarity Report for a specific submission, which is why it carries weight with your marker [3]. If your goal is to understand the risk attached to your actual assignment, a tool that evaluates the actual file is the closer match to the decision that matters.
There is also a practical disagreement problem. Independent detectors frequently produce different scores on the same passage because they were trained on different data and calibrated against different thresholds [2]. Turnitin's own guidance anticipates this by instructing instructors to interpret the report with context and to avoid treating a percentage as a standalone finding [3]. For a student, the operational takeaway is to prefer the detector whose output will actually be seen by the person grading you, and to treat any other score as a rough second opinion [3].
Can a Browser Extension Tell You What Your Professor's Turnitin Report Will Say?
No — not exactly, and the reason is structural rather than a matter of detector quality. The Turnitin report is generated from the file you submit through your course's learning management system, and it is that uploaded artefact, with its formatting, file type, and version, that produces the AI score and similarity matches your instructor reviews [4]. A browser extension never sees that artefact; it sees text rendered on a page [1]. The two pipelines can therefore disagree even when the underlying prose is character-for-character identical [4].
What your instructor actually opens is the Similarity Report inside the LMS, where the AI writing indicator sits alongside match highlights and source links [4]. That combined view is the thing being interpreted, and interpretation is where context enters: Turnitin's institutional guidance directs staff to weigh the indicator against the assignment brief, the drafting process, and any explanation the student provides [4]. A bare percentage from a browser add-on has none of that surrounding evidence attached to it [2].
The version you check also matters more than most students expect. Editing a draft, changing the file format, or submitting a different revision changes what gets analysed, so a score you obtained on an earlier draft is not a prediction of the score on the final one [4]. This is the single strongest argument for checking the actual file you intend to submit, in the format you intend to submit it, rather than checking prose in a browser tab [4].
So the honest answer to the question in the heading is that a browser extension can give you a directional signal and nothing more. If you want to know what your professor's report will say, you need a report generated the same way theirs is — from your real document, with the same AI and similarity views they will see [4]. That is a different product category from a Chrome add-on, and it is the one that answers the question you are actually asking [3][4].
A browser extension is a useful early check, but it will never show you the report sitting in your professor's inbox. If the question you actually care about is "what will my Turnitin AI score and similarity score be on this submission," turnitin0 runs the real check on your actual document and hands back both reports — the same AI writing indicator and similarity view your instructor opens — so you are reading your own results rather than guessing from a page-level score.
※ Turnitin0.com - Actual Turnitin AI Report Cover, Score, Flag And Similarity Summary
FAQ
Is the Pangram AI detector Chrome extension free?
Pangram offers a limited free allowance and paid tiers above it, so the extension itself installs at no cost but scoring volume is capped on the free plan [1]. Check the current plan limits on Pangram's site before relying on it for a long document [1].
Does the extension check the same thing as Turnitin?
No. The extension scores text as it appears in your browser, while Turnitin's AI writing indicator is generated from the file you submit through your course's learning management system, alongside similarity matching [3][4]. They are different artefacts produced by different pipelines [4].
Why did Pangram and Turnitin give me different scores?
Detectors are trained and calibrated differently, and Pangram's own documentation notes that results depend on the model behind the text and on editing [2]. Turnitin also warns that its indicator must be interpreted in context rather than as standalone proof [3].
What does a *% result mean?
On Turnitin, *% is displayed when detection confidence falls below the threshold, so it signals a low-confidence result rather than an exact percentage [3]. It should not be read as either a clean pass or a confirmed finding [3].
Can I lower my AI score by editing after the extension flags me?
Editing genuinely changes what a detector sees, which is why a score from an earlier draft does not predict the score on your final file [4]. Rewrite flagged passages in your own voice and re-check the actual document you intend to submit [2][4].