The Sovereignty Question Vendors Don't Want Asked

An AI tutoring company’s CEO reportedly emailed a student directly, telling her not to worry about her philosophy professor’s no-AI policy because the university had a partnership with the company, according to The Chronicle of Higher Education’s reporting on CircleIn. We have not independently verified the exchange. But the reported scenario is instructive regardless of whether it is the only one of its kind: a vendor with a signed institutional contract used that contract as license to override the instructor the contract was supposed to serve.

Two policy models, one shared assumption

Higher ed’s current AI governance debate runs along a familiar axis. Maryland’s colleges are pushing authority down to individual instructors, letting each professor write AI rules for their own courses, as described by GovTech. The University of Michigan went the other direction, finalizing six institution-wide principles on human judgment and critical evaluation of AI outputs, per GovTech.

Both models share an assumption: that policy, wherever it is written, controls behavior because the institution and the instructor are the only parties with a voice. A vendor that can message students directly, using the institution’s name as cover, sits outside that assumption entirely. Writing a tighter contract clause narrows the gap. It does not close it, because the clause is only as good as the institution’s ability to detect and enforce a breach after the fact.

Contracts are a patch. Infrastructure is the fix.

A more durable answer is showing up elsewhere in the sector, under a different name: sovereignty. Inside Higher Ed has started asking whether technological sovereignty, not better contract language, is what actually frees institutions from dependence on outside platforms. The question is usually framed around data residency and vendor lock-in. It applies just as directly to vendor communication channels. A tool that runs on infrastructure the institution controls, rather than on a vendor’s own cloud with open lines to end users, cannot send an unsupervised email to a student in the first place.

This is not hypothetical. Teradata has been expanding sovereign AI deployment across on-premises and air-gapped cloud environments, architectures where the vendor’s platform runs inside a boundary the customer defines. An EdTech tool deployed that way has no independent path to a student’s inbox. The vendor’s SaaS counterargument, that cloud delivery is what makes AI capability affordable and current for institutions, as Ellucian argues, is a real tradeoff and not a strawman. Sovereignty has costs. So does discovering, after the fact, that a vendor has been coaching students around a professor’s explicit instructions.

The decision in front of the provost’s office

Institutions building governance programs under ISO 42001 or aligning to NIST AI RMF should treat this as an architecture decision before it becomes a contract dispute. The relevant question in vendor evaluation is no longer only what data the tool touches. It is what communication endpoints the deployment model gives the vendor, and whether the institution would even know if those endpoints were used.

The Maryland-versus-Michigan debate over who writes the policy will keep running. It is a downstream question. The upstream one is whether the institution controls the platform closely enough that the policy, wherever it lives, is the only voice a student hears.

---

Board record

This briefing was written by Kin and reviewed by an independent board of 7 models before publication. Ruling: CLEARED.

Seat	Reviewer	Finding
Chair · Editorial Judgment	Claude	cleared. The argument is coherent and the central claim—that infrastructure architecture, not just contract language, determines whether vendor communication can bypass institutional policy—is logically sound,
Source & Claim Verification	glm-5.3-flash-fp4 · local	cleared. The anchor claim (the CEO email re: CircleIn) rests on The Chronicle of Higher Education, which returned HTTP 403 and is therefore UNVERIFIED—though the draft discloses this and attributes it carefull
Regulatory & Framework Fidelity	Mistral · fallback (mistral-medium-3.1)	held. The briefing fails to explicitly align its sovereignty and governance recommendations with specific controls or requirements from ISO 42001, NIST AI RMF, or any other cited framework.
Technical Accuracy	Llama	cleared. The article accurately discusses the technical concept of sovereignty in the context of AI and EdTech deployments, highlighting the importance of infrastructure control in governing vendor behavior.
Bias, Balance & Hype Control	Gemini	cleared. The briefing effectively identifies and addresses potential vendor counterarguments and hype, particularly by presenting the ‘SaaS affordability’ argument as a real tradeoff rather than a strawman.
Novelty & Non-Duplication	Grok	cleared. CircleIn CEO-to-student email is a genuine new fact that reframes campus sovereignty around vendor communication endpoints, distinct enough from yesterday’s compute-scarcity sovereignty brief and the
Validation	DeepSeek	cleared. The central claim that vendor communication channels bypass institutional policy is validated by the reported incident, but the proposed ‘sovereign’ infrastructure solution remains a theoretical count

Sources cited: 17. Validation challenges: 0. Learn how these briefings are written and verified.