A curtain-wall subcontractor’s delegated-design engineer uses an AI drafting tool to lay out anchors. The output looks right. The engineer skims it, seals it and sends it.
The submittal clears review. The problem a real check would have caught shows up at installation. (This scenario is illustrative.)
Who’s responsible? The answer didn’t change because a tool was involved.
Who this applies to
The ethics rules and board guidance below apply to licensed design professionals, such as engineers and architects. Their seal is the official stamp that says they take responsibility for the work.
For a contractor, these rules come into play on design-build, delegated design and temporary-works engineering. Delegated design means a sub’s engineer designs and seals one system, like the curtain wall above.
Mistakes in estimates or schedules are a different question. Those depend on your contract and your contractor’s professional liability (CPL) policy.
AI is already in the design office
Deltek’s latest Clarity study found the share of architecture and engineering firms using AI rose from 53% to 70%. Only 38% reported measurable business benefits.
An ethics board already weighed in
In 2024, the Board of Ethical Review at the National Society of Professional Engineers (NSPE) published Case 24-2. In it, an engineer used open-source AI to draft a report and design documents for a groundwater infrastructure project.
The engineer gave the AI-generated plans a quick once-over and sealed them. The client found misaligned dimensions and missing safety features required by local regulations.
The board’s finding was simple: the tool wasn’t the problem. Using AI drafting tools was “not unethical per se.” Sealing the work without keeping responsible charge was unethical.
Responsible charge means the engineer’s direct control and personal supervision of the work. The board said engineers should challenge AI output the way they would an intern’s work.
Two other findings matter:
- Putting the client’s confidential data into the AI tool required the client’s consent.
- There’s no general duty to disclose AI use unless the contract requires it. Still, the board favored transparency.
This is an advisory opinion, not law. Think of it as guidance, not a ruling. But it shows how a panel of fellow engineers reads the facts.
Licensing boards agree
North Carolina’s engineering board puts it bluntly in its guidelines: “Maintain complete responsibility for the work, regardless of the degree of AI assistance.”
NCEES, the group that writes the engineering licensing exams, says licensed professionals “retain ultimate responsibility for decisions.”
Idaho’s licensing board shared that position with its licensees in 2026. It noted that the position isn’t part of Idaho law.
On the liability side, the AIA Trust, which serves architects, notes that “the standard remains unchanged: negligence is evaluated based on outcomes.”
Lawyers got here first
By June 2026, one tracking database counted 1,058 US cases involving improper AI use in court filings. That’s three years after Mata v. Avianca, the case that made fake citations famous.
The problem has reached construction. In A2K Inc. v. Fluor Corp., a federal case in Texas, a staffing subcontractor admitted using generative AI in a legal brief. That brief allegedly contained a nonexistent quote.
Will insurance cover it?
Design firms carry errors and omissions (E&O) insurance to cover professional mistakes. In October, ENR reported it found no AI exclusions yet on policies for design firms and contractors.
There is one warning. PDI, a wholesale insurance broker, reports that at least one major carrier has added an absolute AI exclusion to E&O. “Absolute” means it is written broadly.
Carriers are also giving advice. AXA XL’s design-professional team has advised against uploading firm designs to ChatGPT. It has also recommended human checks of AI output.
What this means for you
AI doesn’t change the standard of care: the level of care a reasonable professional would use. Using AI isn’t the breach. Signing what nobody checked is.
If your firm seals work, or relies on a sub who does, the human review is the control that matters. Make it real, and make it visible.
If you’re the design-builder, your design contract is with the owner, not your sub. Ask counsel how much of that risk your subcontract passes back.
Do this month
- Write a review rule for AI-assisted licensed work: who checks, what they check and how it’s recorded.
- Get written client consent before client data goes into any AI tool.
- Log AI use on sealed deliverables and delegated-design submittals.
- Ask delegated-design subs to disclose AI use on sealed submittals and name who reviewed it.
Ask your broker
Your policy’s definition of professional services helps decide what’s covered. Put this question to your broker in writing:
“Does our professional or CPL policy define professional services broadly enough to include AI-assisted work? Is any AI endorsement attached or proposed at renewal?”
An endorsement is an add-on that changes the policy’s terms.
Next week
Part 4: what underwriters and sureties now ask about AI, and the documents that answer them.
Part 4 publishes Tuesday, November 10, 2026, at 7:45 a.m. Mountain. Get it by email · Subscribe on LinkedIn · RSS feed
Free resource: Section 7 of our free AI Use Policy for Contractors sets out human review and accountability. Download it as a PDF or Word file. No form to fill out.
Get The Record by email
Each part in your inbox the morning it publishes.
Then the occasional new guide or template for contractors. Unsubscribe in one click.
We confirm your address by email first. Privacy Policy
General information, not legal or insurance advice. Talk to your broker and counsel.
Sources
- NSPE Board of Ethical Review Case 24-2, July 18, 2024
- NCBELS AI guidelines, July 16, 2025
- Idaho PELS letter sharing NCEES position statement, Jan 23, 2026
- AIA Trust: Trends in Professional Liability, Jan 16, 2026
- UNC Law Library: Handling Hallucinations, June 8, 2026
- Jones Walker: Fake AI citations in a construction lawsuit, June 24, 2026
- IRONPROS: Deltek Clarity A&E findings, Aug 15, 2026
- ENR: AI exclusions won’t soon be popping up in designer liability policies, Oct 9, 2026
- PDI: AI Tools and Your E&O Policy, Apr 22, 2026
- AXA XL: AI risks for design professionals, Feb 21, 2024
About The Record
The Record is written by Jared Smith, founder and CEO of SymphonIQ, which builds governed, single-tenant AI for engineering and construction.