Semafor finds AI-written guest op-eds in the NYT, WSJ and Washington Post
Dartmouth provost’s case becomes a campus scandal
Confirmed
The takeaway
On Aug 26, 2026 Semafor ran a month of guest op-eds from the New York Times, Wall Street Journal and Washington Post (310 pieces) through the AI detector Pangram.
Status
- Claim
Confirmed
- Our reporting
- High confidence
- Importance
- 2 of 5
- Last verified
- 2 October 2026
Your AI and this story
- GPT-6.2 Astra87 days after its cutoff
- Claude Opus 5.557 days after its cutoff
- Gemini 4 Hydrogen148 days after its cutoff
- Grok 4.787 days after its cutoff
None of these four assistants can know about it. The closest, Claude Opus 5.5, stops 57 days before it.
Key facts
- Semafor (Aug 26): 310 guest opinion pieces in NYT, WSJ and WaPo over one month; 10 scored 80%+ AI-written on Pangram, ~40 more partly AI-generated
- Druckenmiller’s WSJ bond-market op-ed scored 100% AI; he said: ‘I write everything using AI now for the same reason I use a calculator when I do math problems’
- Also flagged (per press): an Aug 2 NYT guest essay by former CISA director Jen Easterly; the Times declined to comment
- Newspaper policies: the NYT bars AI in ‘developing and drafting guest essays’; the WaPo requires contributors to confirm pieces weren’t ‘created or manipulated with artificial intelligence’; WSJ’s Paul Gigot said what matters is whether a piece reflects ‘an author’s original argument’
- Schnell’s WaPo column ‘Universities are fighting AI cheating. But there’s a deeper problem.’ scored 100% AI. He said he used ChatGPT ‘to refine a few arguments, check for grammatical errors’
- The Dartmouth (student newspaper, Sept 2026) with Pangram 4.0: all of Schnell’s pre-2022 writing scored ‘100% human-written’; nine 2026 works (op-eds, two math papers, emails) had a median 96% AI score
- Sept 25: Schnell apologized for not consistently disclosing AI use, and President Sian Leah Beilock set up an independent committee to review it. The campus AAUP chapter circulated a petition, and a student editorial called for his firing (Inside Higher Ed)
- Caveat: detector-based findings; Semafor noted research showing detectors can be biased against non-native English writers
What happened
Semafor used Pangram to measure how much AI writing appears in the most prestigious opinion pages in the US. About 3% of a month’s guest op-eds scored as essentially machine-written, and over 10% more as partly machine-written. The Dartmouth then examined one flagged author, the provost of Dartmouth College, and found a sharp change in his writing after ChatGPT’s release. The irony that his flagged column was about AI cheating turned the finding into a national higher-education story.
Why it matters
This is a concrete measurement of undisclosed AI ghostwriting in elite public discourse, and an early case of senior institutional leaders being held to the AI-disclosure rules they impose on students. It also shows detectors like Pangram becoming de facto enforcement tools for newsrooms and universities. In October Dartmouth authorized faculty to use Pangram on student work.
Sources
5 sources from 3 sites. Numbers match the chips in the text.
5 sources: 4 press, 1 reaction
Press
- Semafor: Exclusive: AI writing has already begun to appear on the opinion pagessemafor.com, press
- The Dartmouth: Op-eds, academic articles by Provost Santiago Schnell flagged as ‘AI-written’ by Pangramthedartmouth.com, press
- Inside Higher Ed: Dartmouth provost’s ‘AI dependency’ sparks backlash (Sept 29)insidehighered.com, press
- The Dartmouth: Dean of faculty authorizes professors to use Pangram to review student work (Oct 2026)thedartmouth.com, press
Reactions
- Inside Higher Ed (John Warner): Dartmouth’s provost has stopped ‘writing’insidehighered.com, discussion
Changes
- Filed (Zvi AI #188)