Dartmouth Provost Resigns After Undisclosed AI Use Contradicts His Own Warnings

Santiago Schnell wrote a high-profile Washington Post op-ed warning that universities are failing to grasp the deeper problems of AI cheating. This week, the Dartmouth College provost discovered those depths firsthand when he resigned after an advisory committee found he failed to disclose significant artificial intelligence use in an academic paper and that same op-ed, despite publisher requirements.

An investigation revealed that Schnell used tools like Claude to revise papers, edit code, and generate figures, while dodging full cooperation by withholding personal account logs and refusing to meet with the committee. Our read is that Schnell fell victim to the exact grey zone he wrote about: treating LLMs as mere editorial partners while failing to navigate the strict, shifting transparency mandates demanded of academic leaders.

Schnell has long argued that distinguishing strictly between human and AI writing is a false binary, insisting that authorship remains human as long as the writer takes full responsibility. In a September paper on AI detection, he detailed how he used various models to brainstorm structures, check literature, and critique drafts.

Yet while he defended this hybrid method in theory, he repeatedly failed to disclose his chatbot assistance when submitting formal work to journals.

An advisory committee report noted that his logs even showed him asking Claude if an article contained many signatures of AI writing, raising questions about whether he tried to obscure machine assistance.

Dartmouth itself continues to barrel ahead into generative tools, hosting anniversary events and pushing for more campus integration under President Sian Leah Beilock.

But Beilock emphasized that leaders must model transparency and human judgment in a fast-changing environment.

Meanwhile, student surveys reveal mounting campus anxiety, with seventy percent of college students stating that take-home work has become unfair due to AI, and thirty-nine percent feeling pressured to use it because peers do.

Schnell may have argued that contribution and authorship are different roles, but his downfall shows that in elite academia, failing to disclose the helper in the room remains a career-ending offense.