AI Regulation Tracker / United States, professional-body ethics opinion
NSPE Board of Ethical Review Sets the Rules for a Licensed Engineer Using AI
In Case 24-2, the National Society of Professional Engineers Board of Ethical Review applied the NSPE Code of Ethics to a professional engineer who used open-source AI to draft a report and AI-assisted tools to draft design documents. The verdict: AI is a tool, the licensed engineer keeps personal responsibility, and the AI output gets the same scrutiny as human work before a seal goes on it. Issued July 18, 2024.
Most of the AI-and-professionals story right now is statutes and court sanctions. This one is quieter and, if you advise licensed engineers, more useful. It is an ethics opinion from the body that writes the code most state boards lean on, and it tells you exactly where the line sits when a professional engineer puts AI into the workflow and then signs and seals the result.
The date matters, so I will be straight about it. Case 24-2 was issued July 18, 2024. It is not breaking news. It is settled guidance that has been sitting on the books while the rest of the professional world catches up to the same questions, and it reads like it will hold for years. I am covering it now because the reasoning is the clearest professional-body statement I have seen on what a licensed operator owes when AI does part of the work.
The facts, in plain terms
Engineer A is an environmental engineer with several years of experience and a PE license. A client hired the engineer to write a report on a contaminant and to produce design documents for groundwater infrastructure. The engineer was strong technically but shaky on writing, and the mentor who used to review drafts had just retired. So the engineer used open-source AI to generate a first draft of the report, and AI-assisted drafting tools to generate a preliminary design.
For the report, the engineer did the work: a thorough review, cross-checking facts against professional journals, checking phrasing against search engines, and personalizing the wording. For the design documents, the engineer did only a cursory, high-level review. The client caught the difference. The report read cleanly. The AI-generated plans had misaligned dimensions and left out safety features that local regulations required.
Report writing: ethical only if you actually do the review
The Board held that using AI to help write the report was not deceptive and did not, by itself, breach competence. The engineer was qualified, kept the document under their direction and control by reviewing and adjusting it, and did not misrepresent anything. That is the part firms should notice: the BER is comfortable with AI as a drafting aid when a competent professional genuinely verifies the output.
But two things still tripped the engineer up. First, uploading the client's private information into a public open-source AI tool without the client's consent was, in the Board's words, tantamount to placing that information in the public domain, which cuts against the duty to protect confidential information. Second, the AI-generated report did not include citations to the technical sources it drew on, and the Code requires giving credit where it is due. So even the report use was, in the end, less than fully ethical.
The design documents: where Responsible Charge broke
This is the core of the ruling and the part with real operational bite. The engineer sealed design documents after only a high-level look, and the AI had produced errors a proper review would have caught. The Board tied this straight back to two anchors of the Code. The safety errors ran against Fundamental Canon I.1, to hold paramount the safety, health, and welfare of the public. And by sealing plans they had not meaningfully supervised, the engineer failed to keep Responsible Charge, which the Board treated as a violation of the duty to conform with state registration laws under Code section III.8.a.
The Board was blunt about the standard. In its words, "AI-generated technical work requires at least the same level of scrutiny as human-created work." It reached back to a 1998 opinion for the governing principle that technology "must never be a replacement of a substitute for engineering judgment," and framed responsible AI use like directing an intern: you set the constraints, you challenge the output, and you accept it only after you have satisfied yourself it meets professional standards. Reviewing a drawing after the fact, without being engaged in the design, does not count as Responsible Charge.
What about disclosure
The engineer did not tell the client AI was involved. The Board's answer here is careful, and worth stating precisely so no one oversells it. There is no per se duty to disclose AI use to a client, much like you would not itemize every piece of software used in a design, unless the engineer's contract requires it. At the same time, the Board noted there is currently no universal guideline mandating AI disclosure in engineering work, and that ethical principles favor transparency when AI plays a substantial role. Its closing recommendation: engineers integrating AI into their practice should "adopt rigorous verification processes and consider disclosing AI involvement when it plays a significant role in the final product."
Why a US professional practice should care
An NSPE ethics opinion is not a court order and it is not a statute. But the NSPE Code of Ethics is adopted or referenced by many US state professional engineering licensing boards, and boards and courts look to BER opinions as persuasive authority on what competent, ethical practice looks like. So the practical read is this. If your engineers use AI, the seal still means a human took Responsible Charge. Cursory review of AI output before sealing is the failure mode the Board just labeled unethical. And putting client data into a public AI tool without consent is its own separate problem, independent of how good the output is.
Questions professionals are asking
Does NSPE Case 24-2 ban engineers from using AI?
No. The Board of Ethical Review treats AI as a tool that is acceptable in engineering practice, the same way it accepted computer-aided design decades ago. What it requires is that a licensed engineer stay competent, keep Responsible Charge, and review AI output at least as carefully as human-created work before relying on it or sealing it. AI cannot be a substitute for engineering judgment.
Is this opinion legally binding?
Not by itself. A BER opinion is an advisory interpretation of the NSPE Code of Ethics, not a statute or a licensing-board order. Its weight comes from the fact that the NSPE Code is adopted or referenced by many US state professional engineering licensing boards, and boards and courts treat BER opinions as persuasive authority on ethical practice.
Where exactly did the engineer go wrong?
Two places. On the report, the engineer uploaded the client's private information into a public open-source AI tool without consent and did not credit technical sources. On the design documents, the engineer did only a cursory review, missed misaligned dimensions and omitted safety features, and sealed the plans without maintaining Responsible Charge, which the Board tied to the duty to hold paramount public safety and to conform with state registration laws.
Do engineers have to tell clients they used AI?
Under this opinion, there is no per se duty to disclose AI use unless the engineer's contract with the client requires it, similar to not itemizing every piece of design software. The Board noted there is no universal AI-disclosure guideline yet, but recommended adopting rigorous verification processes and considering disclosure when AI plays a significant role in the final product.
When was Case 24-2 issued?
The NSPE Board of Ethical Review issued Case 24-2 on July 18, 2024. It is standing guidance rather than a recent development, and it interprets the existing NSPE Code of Ethics for the AI context.
RELATED BRIEFINGS
Browse the full AI Regulation News tracker
Informational analysis for working professionals, not legal advice. NSPE BER opinions are advisory; confirm how the NSPE Code and your state licensing board apply to your situation with qualified counsel.