ABET updated its AI policy page on July 9, 2026. The change was not on the front page of any news site. No press release accompanied it. But for anyone responsible for engineering program accreditation, the signal is clear: AI is now a formally recognized part of the accreditation ecosystem, and the conversation about how programs use it is no longer speculative.
For Canadian programs, the question is not whether ABET’s policy applies to you. The question is what it signals about the direction accreditation is heading — and whether your evidence infrastructure is ready for it.
What ABET Changed
ABET’s “Accreditation and Artificial Intelligence Technologies” page was formally updated on July 9, 2026. The page articulates ABET’s official position on how AI fits into the accreditation process — both for programs preparing their self-studies and for ABET’s own review operations.
The policy covers three areas, each with clear boundaries:
1. Accreditation review preparation
ABET explicitly permits programs to use AI technologies for gathering and synthesizing information in preparation for accreditation reviews, for supporting the collection and analysis of assessment data, and for assisting in the development and organization of supporting materials.
There are conditions. Human oversight is mandatory. All submitted information must be verified for accuracy by qualified personnel. AI must not alter, manipulate, or misrepresent collected data or documented evidence. The program must acknowledge its use of AI technologies, and the institution must attest to the accuracy of the narrative and data presented.
2. The accreditation review itself
On ABET’s side, the policy states that AI technologies shall not replace human judgement in any aspect of the accreditation evaluation process. All materials and deliberations remain subject to ABET’s confidentiality policies. This is a governance commitment: the people reviewing your program are humans, not algorithms.
3. Internal administrative processes
ABET requires that any AI tools used for its own administrative processes be developed using verified data, evaluated before implementation for accuracy and bias, and continuously monitored following deployment. This section signals that ABET is building AI capabilities internally — and doing so with the same rigor it expects from programs.
The definitions section is worth noting. ABET defines “human judgement” as “the deliberate exercise of reasoning and decision-making by individuals, informed by knowledge, experience, contextual awareness, ethical considerations and personal accountability.” That is a high bar, and it is deliberately placed between the program and the AI tool.
Three Takeaways from the Policy
Reading between the lines, three signals stand out for programs that are thinking about their accreditation infrastructure:
Signal 1: AI is no longer optional
Policies do not appear in a vacuum. ABET did not write this page because AI is a fringe topic. It wrote it because programs are already using AI — some effectively, some recklessly — and the accreditation body needed to establish guardrails. The mere existence of the policy tells you that AI use is widespread enough to require formal guidance.
The programs that benefit most from this moment are the ones that have already built AI-ready data infrastructure. When the policy says AI can support “collection, analysis and evaluation of assessment data,” it assumes that data is structured, accessible, and verifiable. If your assessment data lives in disconnected spreadsheets across twelve faculty desktops, AI cannot help you. If it lives in a connected system with clear source links and audit trails, AI can dramatically reduce the time between raw data and actionable insight.
Signal 2: Human oversight is the non-negotiable
The policy repeats the same principle in three different sections: AI must not replace human judgement. Every output must be verified by qualified personnel. The institution must attest to accuracy.
This is not anti-AI language. It is pro-accountability language. The accreditation process carries professional and reputational weight. The data you submit determines whether your graduates can practice engineering. AI is permitted as a tool — but the responsibility for what gets submitted rests entirely with the program.
In practice, this means the AI tool you choose matters. A general-purpose chatbot cannot provide the source transparency, correction logging, or confidence scoring that makes AI output reviewable. A purpose-built tool that grounds every AI output in your own evidence, links back to source documents, and records every human correction creates the audit trail that satisfies both the policy and your own quality standards.
Signal 3: The conversation is global
ABET is a US-based organization. But under the Washington Accord — the mutual recognition agreement between engineering accreditation bodies in 17 countries, including Canada — developments at ABET tend to propagate to peer organizations. CEAB monitors ABET’s criteria changes, policy updates, and assessment practices. When ABET formalizes something, CEAB takes notice.
This does not mean CEAB will copy ABET’s AI policy. The Canadian context is different: stronger data privacy expectations, different institutional structures, and the PIPEDA framework that governs how personal data — including student assessment data — can be processed and stored. But the direction of travel is clear. Accreditation bodies worldwide are working out how AI fits into quality assurance. The programs that build their evidence infrastructure now will be ahead of the curve when CEAB eventually issues its own guidance.
The Canadian Gap: CEAB Has No AI Policy Yet
As of July 2026, Engineers Canada and CEAB have not published an AI policy for accreditation. That is not a gap in ambition. It is a gap in timing.
CEAB is currently focused on two active initiatives. The “Realizing Futures of Engineering Accreditation” strategic program is consulting stakeholders on the direction of Canadian accreditation. The faculty licensure consultation — which closed in early August 2026 — examined whether changes to accreditation criteria should align with proposed changes to professional engineering licensure requirements.
AI in accreditation is likely to enter the CEAB conversation in the next criteria revision cycle, which is planned for Spring 2027. When it does, the questions will be familiar: How do programs use AI responsibly? How is human judgement preserved? What data governance standards apply?
The programs that will be in the strongest position are the ones that have already answered those questions in practice. Not in theory. In practice. With structured evidence, verifiable data trails, and AI tools that respect Canadian data residency requirements.
What Canadian Programs Should Do Now
You do not need to wait for CEAB to issue guidance. Here are four actions that position your program well regardless of what the policy landscape looks like in 2027:
1. Build connected data infrastructure
AI is only as good as the data it works on. If your evidence is scattered across spreadsheets, shared drives, and individual faculty desktops, no amount of AI will fix that. The foundational work is connecting your data: courses to indicators, assessments to outcomes, improvement actions to evidence. Once that structure exists, AI can operate on it reliably. If it does not exist, AI will amplify the chaos.
2. Ensure Canadian data residency
PIPEDA and provincial privacy laws apply to student assessment data, faculty information, and institutional program records. When you evaluate AI tools for accreditation, the hosting location matters. Tools that process Canadian program data on US servers create compliance questions that your IT department and procurement office will not let slide. Choose tools with Canadian data residency from the start, not as an afterthought.
3. Adopt tools that show their work
The ABET policy requires human verification of all AI output. The practical implication is that the AI tool must show you what it did, why it did it, and where it got its information. Look for source transparency, confidence scores, and correction logging. If a tool cannot explain how it arrived at a mapping or a gap finding, you cannot verify it. If you cannot verify it, you should not use it.
4. Monitor the regulatory landscape
Engineers Canada’s consultation processes, ABET’s criteria updates, and the broader AI governance conversation in Canadian higher education are all worth watching. The Summer 2026 period is when coordinators have the most bandwidth to evaluate tools and plan for the next accreditation cycle. Use it.
How This Fits Together
ABET’s policy update is a marker, not a mandate. It tells you where the accreditation community is heading and gives you a reference point for making decisions now. For Canadian programs, the path is clear: build connected evidence infrastructure, ensure your data stays in Canada, adopt AI tools that support — not replace — human judgement, and stay informed about where CEAB is going.
The programs that act on this now will find themselves in a strong position when CEAB’s next criteria cycle arrives. The evidence will be connected. The AI will be working for them, not against them. The visiting team will see a program that understands the future of accreditation and has already prepared for it.
The rest will be scrambling. As they always do.
This post references ABET’s “Accreditation and Artificial Intelligence Technologies” page, updated July 9, 2026, available at abet.org/accreditation/ai-policy/. CEAB has not yet published an equivalent AI policy. ABET’s policy serves as a practical reference for CEAB programs under the Washington Accord framework.