Plagiarism and AI at Penn State: the XF grade and how the X comes off

The short answer

Penn State can mark a failure on your transcript as academic misconduct — but unlike most universities that do this, the mark is time-limited and can come off early. Knowing that changes what you ask for at the committee stage.

It is also the university with the most explicit published instruction in this whole series about what a committee may not treat as evidence.

The XF grade, and its time limit

Senate Policy 47-20 defines it:

“The XF grade represents a course failure and is only determined through the academic integrity process. The XF grade symbol has the same grade-point equivalency as a failing grade.”

It is not something an instructor can hand out. The procedures document, G-9, sets three conditions at once:

“For egregious violations or multiple violations in the same course, Academic Integrity Committees may assign an XF transcript notation, but only when the instructor asks for an academic sanction that would result in an ‘F’ grade in the course. The Academic Integrity Committee determines the duration of the transcript notation (up to two years), any conditions required for earlier removal, and notifies Student Accountability.”

Three things follow, and they are all useful:

• The X is capped at two years.Compare Texas A&M, where an equivalent mark’s underlying F is permanent, or UCF, where a second finding makes the notation permanent outright.
• Early removal is contemplated — the committee sets the conditions, which means there is something to ask about rather than simply accept.
• The X can be appealed on its own.The Sanctioning Guidelines note that G-9 “provides a mechanism for students to appeal the ‘X’ portion of the ‘XF’ grade and/or program removal.” You can accept the failing grade and still contest the notation.

A committee can also remove you from a degree programme entirely for “egregious violations or multiple violations in courses required for the completion of a degree program”, and can recommend university-wide sanctions — formal warning, conduct probation, suspension or expulsion.

The published severity ladder is short: minor — redo or reduced grade, with or without an integrity warning; moderate — 0% on the assessment, with or without a course-grade reduction; major— a course grade of F or XF. An integrity warning “has no impact on the student’s academic progress” and is aimed at cases that are “both minor and unintentional”.

Severity is judged on your intent where it can be established, your familiarity with the expectations and the instruction you have had, how much of the assignment was affected, what the assignment was worth, the impact on others, and how long you have been a student. If you are early in your degree and nobody had taught you the referencing rules yet, that is a listed factor — say so.

The consequences nobody mentions

These are in the policy and they routinely surprise students more than the grade does:

• “A student found in violation of academic integrity is ineligible for appointment to the Dean’s List for the semester […] during which the violation occurred.”
• Students with an academic sanction in a course may not elect Pass/Fail for that course.
• Grade forgiveness cannot be applied to a course where you received an academic sanction for an integrity violation.

That last one matters: the usual route for repairing a bad grade is closed for this particular bad grade. And dropping the course does not end the process either.

On the record itself, there is a real protection worth knowing: “Student Accountability does not disclose academic misconduct records unless those records include Administrative Probation or above”. Most findings do not follow you to an external enquirer.

Detector scores are not evidence

If you are ever accused at Penn State on the strength of a percentage, print this. It is from “Guidance Around Faculty Use of AI Detectors”, published on Penn State’s academic integrity site:

“Faculty are discouraged from using AI detectors due to the unreliability of these tools, their biases, and the risk of false positives. Many of the linguistic markers that AI generators use for detection are also used by human authors, leading to an unverifiable pattern of false positive rates. […] AI detectors offer probabilistic assessments that cannot be validated against original sources, which makes them distinct from traditional plagiarism detectors.”

On Turnitin specifically:

“Although the Turnitin platform is available to Penn State faculty, the Turnitin plagiarism detector is not suitable for the detection of AI use. Turnitin sells a different AI detector as an add-on. However, like other detectors, it does not yet meet the accuracy/reliability criteria for determinative use in academic integrity claims, and it has not been purchased.”

And the instruction to the people who decide cases:

“While detectors may help start a conversation with a student about AI use, AI integrity committees should not consider AI detector scores as evidence.”

Two honest caveats. The document is titled “Draft v1.3” and prints no date, though its file path places it at January 2026 — so treat it as current guidance rather than settled policy, and ask your committee to confirm it applies. And “discouraged” is not “forbidden”: an instructor can still run a detector. What the guidance says is that its output should not carry weight in the decision.

Penn State also reports its own research: “Faculty did not find AI detection tools to be accurate and were skeptical about their reliability. Students had reservations about instructors’ use of AI for grading, were worried about allegations for the improper use of GenAI, and felt unsure their instructors would believe their denials of use.” That last clause describes the actual harm well.

The same reasoning applies to our own tool, and we say so wherever this comes up: our published error rates are measured on English and German corpora, and no detector produces proof. See how accurate AI detectors really are and what a Turnitin score actually means.

What Penn State looks at instead

Its FAQ for educators answers the detection question without naming a single tool. The list is entirely behavioural, and it tells you exactly what to be ready for:

“The most effective way is the most obvious – when you suspect a violation, talk with the student. It is especially noteworthy when the student cannot answer reasonable questions about the work they’ve submitted.”

The rest of the list: AI tools self-identifying in the text; fabricated citations, references, names, quotations or interviews; information from outside the permitted source set; the same examples or errors recurring across students’ work; unusually short completion times; and “significant changes in the student’s ‘voice’ within or across assignments”.

The practical reading: be able to talk about your own work, verify every citation exists, and do not let your writing style lurch between assignments because you leaned on a tool for one of them. Our guides on proving you wrote it yourself, verifying AI citations and what to do when you are accused cover each of those.

If a detector score is what started this, our guide to an AI detector flagging your thesis sets out what that number does and does not establish on its own.

The Sourced Tool citation

Where AI use is permitted, Penn State publishes a more concrete form than any other university in this series:

“Students must include an in-text citation directly after any sentencecontaining information that has been re-worded, paraphrased, copied, or created by a generative AI tool. Students must put copied text within quotation marks and must indicate when other media (e.g., images, graphics, etc.) has been copied or generated. Students must also list the AI tools they’ve used as a ‘Sourced Tool’ at the end of the paper or project.”

Note “re-worded” and “paraphrased” sit alongside “created”. Having a tool rephrase your own sentence triggers the citation just as much as having it write a new one. Sentence-level, not a blanket note at the end — plus the Sourced Tool list.

There is a related restriction in the same document: “Students may not use generative AI tools to complete multiple-choice, matching, fill-in the blank, open-ended, or essay exam questions” — exams are out even where coursework is in.

Our AI disclosure statement guide covers the general shape, but at Penn State follow the syllabus form exactly — it is more prescriptive than most.

Grammarly is in scope

This catches people, so it is worth its own heading. Penn State’s syllabus language sweeps in tools students do not think of as AI:

“This policy applies to the use of all generative AI tools, including Grammarly, Copilot, and other artificially intelligent tools provided by the university.”

And in the variant that permits AI only for specific constructive processes — brainstorming, outlining, improving grammar — the exclusion is explicit: other uses are prohibited, “including the use of AI tools to create content for coursework, including the AI tools embedded in Grammarly.”

So “the university gave me this tool” is not a defence, and neither is “it only fixed my grammar” if the tool did more than that. Check which of the four syllabus variants your course uses.

Unintentional counts

Penn State’s operative definition puts this in the first line:

“An academic integrity violation is an intentional, unintentional, or attempted violation of course or assessment policies to gain an academic advantage or to advantage or disadvantage another student academically.”

You do not have to have meant it. Intent shows up in severity and sanction, not in whether a violation exists. And note “attempted” — the offence can be complete without any advantage actually gained.

The five categories cover unauthorised collaboration and “accessing or using unauthorized or prohibited materials, information, tools, technologies, or study aids”; misrepresenting another’s words or ideas without attribution; providing false information including fabricated data or sources; reusing your own work for credit more than once without permission; and retaining or disseminating course material without permission.

Senate Policy 49-20 itself does not mention AI at all. It arrives one level down — in the Sanctioning Guidelines, whose Misrepresentation examples include “submitting work created by generative technology without attribution”, and in the required syllabus statement. The policy also puts the duty on your instructor: “At the beginning of each course, it is the responsibility of the instructor to provide students with a statement clarifying the application of University and College academic integrity policies to that course.”

Your AI Studio prompts

A small point, but a reassuring one that few universities address at all. Of its own AI service, Penn State says:

“Penn State IT staff do not actively monitor your individual prompts or conversations. […] Prompts and conversations will not be provided in connection with academic integrity investigations.”

Using the university’s own AI service does not create an evidence trail that will be turned against you. That is a genuine argument for using the sanctioned tool rather than a personal subscription.

On similarity, Penn State runs Turnitin Feedback Studio both inside Canvas and as a standalone web app at turnitin.psu.edu, where students self-enrol with an instructor-provided key — so a self-check route exists, but it depends on your instructor setting up the class. Whether submissions enter a comparison repository, and any opt-out, is not published.

Before you submit

• Find your course’s AI variant. Penn State publishes four, from fully permitted to fully prohibited.
• Cite AI at sentence level, plus a Sourced Tool list. A single note at the end is not the published form.
• Count Grammarly. It is named in the policy language.
• Be able to discuss your own work. That is the first thing the FAQ tells instructors to test.
• If you get an XF, ask about duration and early removal. Both are set by the committee, and the X is appealable separately.

One gap worth knowing: Penn State does not state a standard of proof at university level. Most US institutions use preponderance of the evidence; Penn State does not publish one, so ask what standard your committee is applying.

If you want a read on your own text before you hand it in, our AI check is free up to 1,500 words, with no account and no name.

Sources

• Penn State: AAPPM G-9: Academic Integrity — revised by ACUE, 14 December 2022; the XF rules and the operative definition.
• Penn State: Guidance Around Faculty Use of AI Detectors — titled Draft v1.3; no date printed, file path 2026-01.
• Penn State: Sanctioning Guidelines for Violations of Academic Integrity — updated 7 February 2023.
• Penn State: Language to support course policies on students’ use of generative AI tools — the Sourced Tool citation form; undated.

This page is orientation, not legal advice. What binds you is your course syllabus, Senate Policy 49-20 and AAPPM G-9.