Dean Le BlancCollege faculty member. Still grades every week.
Most coverage read the Dartmouth story as a provost who got caught. Read it instead as a campus finding out that its AI rules only ran in one direction.
Dartmouth’s coursework guidelines are clear about students. No generative AI on submitted work unless the instructor permits it, and any permitted use gets acknowledged under the honor policy. Then a detector put a median score of 96 percent “AI-written” on nine pieces the provost published this year, and the president named the gap herself: “we need role-appropriate rules and policies for this technology.”
Role appropriate. That includes you. The syllabus you handed out in August almost certainly tells students what they may do with AI. Ask whether it says what you do with it: the feedback you drafted, the quiz bank, the rubric wording, the recommendation letter.
Students are not waiting for a committee to settle that. They read the same news you did. The first time one of them asks, the answer should already be in writing.
The first flagged paper of the term. If a detector or a hunch has put one paper in your pile on hold, read Dartmouth’s new guideline before you do anything else: faculty must communicate with the student before detection results are used in “penalizing or otherwise affecting a student’s grade.” It is a good rule even where it is not the rule.
AI marking is a policy question now, not a rumor. The Guardian reported on September 27 that Western Sydney, Newcastle, Deakin, RMIT and Adelaide all allow staff some use of AI in marking or feedback, each with conditions. Deakin says staff can use it to “improve efficiency and support assessment and administrative activities” but not to “assign grades.” Newcastle gives students the right to opt out. The University of New South Wales, Melbourne and Sydney draw the line at using generative AI in marking at all.
Ask yourself the question your students will ask. If you used AI anywhere near feedback this term, would your students know? Armin Alimardani, a senior lecturer at Western Sydney, has a name for what happens when the checking slips: verification drift. A few AI generated assessments look fine, the overworked human stops being as vigilant, and hallucinations slip through as facts.
Pangram. You can get it today. Whether you should is a separate question.
Pangram is the AI detector Dartmouth just named as the only one its faculty are authorized to use, “because it has shown the most reliability of available detectors.” It is also the tool The Dartmouth and Semafor used in the provost reporting.
Can you get it? Yes, and that is unusual for this section. Pangram’s own pricing page lists a free tier that scans up to 2,000 words a day with no payment method, and an Individual plan at $20 a month billed annually for up to 300,000 words a month.
Here is the honest part. A free personal account is not the same thing as institutional permission. Dartmouth required student work to be de-identified before submission, under FERPA, and it is still only “evaluating” an institutional subscription. If your campus has not said how student work may be uploaded to a detector, pasting a paper into your personal account is the step to think hardest about.
And no detector answers the question you actually need answered. Dartmouth’s own standard is a “constellation” of information beyond the result.
Verdict: worth knowing, not worth using alone. Do not bother running student work through a personal account until your institution says how.
Instructure, the company behind Canvas, published its first EdTech Top 40 for higher education this week. It covers nearly 19.5 million U.S. higher education users from September 2025 to April 2026, and it found that no dedicated AI tool ranks among the top 100 tools integrated into the learning environment. Google Gemini, the most widely adopted dedicated AI tool in the analysis, reached nearly 88,000 users across 211 institutions.
Read the caveat before you read the headline. The analysis counts tools formally integrated into the LMS, not the consumer AI apps that students and faculty open on their own.
Which is exactly the point. The AI in your course is mostly not in your LMS. It is in a browser tab nobody counted. You cannot govern that with an integration. You can only govern it with a sentence everyone can read.
Write your own AI line. One course, by hand, fifteen minutes.
One. List what you actually used AI for in this course so far this term. Be literal. Drafting announcements. Generating practice questions. Rewording a rubric. A first pass on feedback. If the answer is nothing, write that down too, because that is also a line students deserve to read.
Two. Mark each one: would a student be surprised to learn it? The surprising ones are the ones that need to be in writing first.
Three. Write the line in the same shape you ask of students. What you use it for, in verbs. What you never use it for, in one clause. How students will know when it was used.
Four. Put it in the syllabus directly under the student AI policy, and say it out loud in class on Monday. Two minutes.
That is your line. Notice that it is short, and notice that writing it changed what you are comfortable doing.
Now look at your other sections. Each line has to match what you actually do in that course, the student policy above it has to agree with it, and both go stale the next time you adopt a tool. Keeping the two in step across every syllabus you teach is the part that quietly slides by midterm. That specific job is what the Syllabus Generator does, working from your Course Record, so the student policy and your own line stay matched in every course and every section.
And write the first one by hand anyway. A line you wrote yourself is the one you will keep.
If a detector is part of your course, today’s free session builds the rest of the case into one assignment: The Constellation File.
- Friday: Dartmouth Just Authorized an AI Detector. Its Own Rules Say the Score Is Not the Case. What the seven guidelines ask of faculty, and the sentence in guideline six most coverage skipped.
- Friday: The Constellation File, a free working session. Your detector line, three points of evidence, and the conversation written before anyone is accused.
Your syllabus this fall tells students where the line is.
The next question on campus is where the line is for the person holding the red pen. The colleagues who answer it in writing first will not have to answer it in a meeting later.
Forward this to the colleague who chairs your academic integrity committee.
Harvest Kernel for the long teaching sessions. Tuesdays faculty, Thursdays students.
The Daily Seed for the short ones.