Scoring basis, stated before the table: these six routes are not compared on quality of output, because a competent editor and a competent writing adviser both improve a text. They are compared on the four things an institution is accountable for — who pays, whether you can see it, whether disclosure happens, and whether a scope limit can be enforced.
| Route | Who pays | Institutional visibility | Disclosure | Scope limit enforceable? |
|---|---|---|---|---|
| 1. Supervisor and committee | Institution | Complete | Not required — it is the expected route | Yes |
| 2. Academic writing centre | Institution | Complete | Generally not required | Yes |
| 3. Institutional platform | Institution | Complete, and it produces a record | Per your policy | Yes — configurable |
| 4. Approved external copy editor | 🔴 The student | Partial — you know it happened | Required, in the document | Partly — by rule and evidence |
| 5. Unapproved paid editing | 🔴 The student | None | None | No |
| 6. Consumer AI tool | 🔴 The student | None | Per policy, if the student knows it | No |
Read down the “who pays” column. Rows 4 to 6 are the same financial arrangement: the institution has stopped providing, and the student has continued buying. The only variable across those three rows is how much you can see.
The rule that gets this right
The University of Toronto’s School of Graduate Studies publishes the most complete institutional position on external editing we have found, and it is worth reading in full because it solves the problem as an escalation ladder rather than as a prohibition.
Its starting premise: the University “expects a thesis submitted to satisfy degree requirements at the graduate level is the student’s work, completed under the guidance of the supervisor and supervisory committee.” And its reasoning for why the drafting matters is pedagogical rather than punitive — “the iterative process of drafting and revising scholarly work deepens students’ engagement with their research, encourages novel contributions, and facilitates the overall research process.”
Then the ladder. The supervisor and supervisory committee “serve as the primary resource in the development and refinement of writing skills,” with additional support available from the Graduate Centre for Academic Communication. Only “if the student and the committee agree that the thesis has ongoing stylistic issues that have not been resolved through available institutional resources” does external editing enter the conversation at all.
And then the sentence that should appear in every business case for writing support:
“It should be acknowledged that engaging with a copy editor for the review of a graduate thesis can be a significant financial burden.”
That is the institution naming the equity problem in its own guidance, unprompted. It is the strongest available argument for provision, and it comes from a graduate school rather than from a vendor.
The four controls, and why each one is load-bearing
Where Toronto does permit external editing, four conditions attach. Copy them.
- Prior approval with a strong rationale, from the supervisor and supervisory committee. This converts a private transaction into a supervised decision, and it is the control that makes the other three possible.
- A hard scope limit: “the intellectual content of the thesis cannot be modified through the external editing process.” This is the line every policy gestures at, expressed in one enforceable sentence.
- Disclosure in the document: the student “must clearly identify the contributions of an external copy editor to their work in the thesis document.” Not on a form filed elsewhere — in the thesis, where an examiner will see it.
- An evidential fallback: students “may also be requested to provide the supervisory committee with copies of the thesis before and after professional editing to demonstrate the role of the copy editor.”
Control four is the one most institutions omit and the only one that makes the others verifiable. Without a before-and-after, “the intellectual content was not modified” is a claim nobody can test. With it, the question resolves in twenty minutes. And the mere existence of the power changes behaviour without ever being used, which is the ideal state for a rule of this kind.
Toronto’s closing requirement is the correct backstop: ultimately the student must provide sufficient evidence that they themselves have met the criteria of the degree. That is the same accountability standard that should govern AI assistance, which is why the two rules belong in one framework rather than two — the reasoning is in how AI rules should differ for a doctoral thesis, and the drafting components sit in what a university AI policy should include.
The three routes you cannot see
Row 5 — unapproved paid editing
A market exists, it is easy to reach, and it is not uniform. At one end sit qualified academic editors who work strictly to a language brief and would decline to touch an argument. At the other end sit services that will rewrite a chapter and describe it as editing. The student choosing between them is often the least equipped person in the transaction to tell which is which, and price is not a reliable signal.
The institutional problem is not that the market exists. It is that when a student buys from it without approval, none of the four controls operates. No rationale, no scope limit, no disclosure, no before-and-after. The work arrives at examination indistinguishable from work that had none of it.
Row 6 — the consumer tool bought privately
The same structure with a lower price and a higher volume. The visibility gap here is the one described in the AI use your institution cannot see, and the point that matters for this comparison is that it is a purchasing decision as much as a behavioural one. A student who cannot afford a copy editor can afford a subscription, and will make that substitution without regarding it as a decision about integrity at all.
Row 4 — approved external editing, honestly assessed
This is a legitimate and sometimes necessary route, and the comparison should not imply otherwise. A candidate writing in a second language, with a supervisor who has confirmed the research is sound and the prose is not, is exactly the case the Toronto rule was written for.
Its limitation is structural rather than ethical: it arrives at the end. A copy editor engaged three weeks before submission cannot develop the writing skill the degree was supposed to build, and cannot fix a structural problem without crossing the intellectual-content line. The cost falls on the student, and the benefit is narrower than either party expects.
The recommendation, and the alternative profile
Recommendation: invest in rows 1 to 3 and regulate row 4 properly. Rows 5 and 6 are not a policy choice you get to make — they are what happens automatically when rows 1 to 3 are exhausted. The lever an institution actually controls is how far out the ladder a student has to climb before leaving your sight.
Row 3 is where the marginal money goes furthest, because it is the only route that scales without a proportional staffing increase and the only one that produces a record as a by-product. The capacity arithmetic is in the writing centre waiting list you cannot staff your way out of.
Alternative profile: a small, research-intensive graduate school with a low supervisor-to-candidate ratio may reasonably conclude that rows 1 and 2 already meet demand. Test that against actual waiting times and against your international cohort specifically, rather than against a general impression.
Neither, if: your problem is detection rather than provision. Support and screening answer different questions, and buying one to solve the other produces a renewal in two years with the original problem intact.
Five things to establish this term
- Do you have a published external editing rule at all? If not, students are making the decision without one, and examiners are judging the results without a standard.
- Does it name a scope limit in one sentence? “Intellectual content cannot be modified” is the sentence.
- Does disclosure go in the thesis, or on a form? It should be in the thesis.
- Can you require a before-and-after? If not, the scope limit is unverifiable.
- What is your current waiting time for institutional writing support? That number is the size of the market you are handing to rows 5 and 6.
Whatever you conclude about provision, measure the effect of what you already run rather than its volume — the method is in how to measure whether a writing support intervention actually worked.
To see what row 3 produces in practice, including the record it leaves for supervisors and examiners, request an institutional evaluation.
Frequently asked questions
Should external copy editors be banned outright?
A ban is hard to enforce and removes the disclosure you would otherwise get. Toronto’s approach — permitted as a last resort, with approval, a scope limit, disclosure and an evidential fallback — is more workable.
Who should be the primary source of writing support?
The supervisor and supervisory committee, with an academic communication centre alongside. Toronto states this explicitly, and external editing sits outside that structure.
What is the scope limit?
That the intellectual content of the thesis cannot be modified through the external editing process. One sentence, and it is the whole rule.
Where should the editor’s contribution be disclosed?
In the thesis document itself, clearly identified, so an examiner sees it without consulting a separate file.
Can we ask for the thesis before and after editing?
Toronto’s guidance provides that students may be requested to supply both, to demonstrate the copy editor’s role. Without that power the scope limit cannot be tested.
Is external editing an equity problem?
Yes, and institutions say so themselves. Toronto’s guidance acknowledges directly that engaging a copy editor can be a significant financial burden.
Are all editing services the same?
No. The range runs from qualified academic editors who work strictly to a language brief to services that rewrite chapters and call it editing. Price does not reliably distinguish them.
How does a consumer AI subscription fit this comparison?
As a cheaper substitute for row 5, with the same absence of visibility, disclosure and scope limit — and a much larger user base because of the price.
Does providing support reduce integrity cases?
It is a reasonable hypothesis and should be tested rather than asserted. Design the evaluation before the intervention, with a comparison group.
Should provision be targeted or universal?
Targeted provision reaches the students who need it most per pound spent; universal provision avoids the labelling problem and reaches those who would not self-identify. Most institutions need a universal baseline with a targeted layer.
What if a supervisor recommends a specific editor?
Handle it as a conflict-of-interest question. The approval decision should not be made by someone with a relationship to the provider.
What is the single cheapest improvement?
Publishing the rule. A clear, findable external editing policy with the four controls costs nothing to write and immediately converts an invisible transaction into a disclosed one.