New research
How AI is being written into — and out of — the insurance behind stamped engineering and design work. What the paperwork says, and what’s being sold on top of it.
Free to read · no sign-up
AI, liability & professional accountability
Defensible AI for licensed AEC practice.
Every AI vendor’s disclaimer says the same thing: verify the output, you’re responsible. Under professional licensure, that’s not just fine print — it’s the law. AECO Shield produces the evidence that a licensed professional stayed in responsible charge of AI-assisted work: the record your E&O carrier, your board, and your own defense will want.
Register interest →What this is, and is not
We produce evidence, not coverage. AECO Shield does not place, price, or guarantee insurance, and it does not certify that an AI tool is safe. It documents that a named licensed professional exercised responsible charge over AI-assisted work — the accountability the seal already carries, made provable. The coverage decision stays with your carrier and broker; the engineering judgment stays with you.
The shift
AI is now routinely used to draft, check, and generate work that a licensed professional seals. The tools disclaim responsibility for it. E&O carriers have noticed: in 2026, AI exclusions began appearing on professional-liability policies, and renewal questionnaires started asking how firms use AI and who reviews its output.
Every major AI tool cascades responsibility back to the user.
Responsible charge is personal and non-delegable — it cannot pass to software, whatever the tool claims.
What used to be silent is now a renewal question, and increasingly an exclusion.
The firms that can show a licensed professional stayed in responsible charge of AI-assisted work are in a different position at renewal than the firms that can only say so.
What the work produces
For each sealed deliverable that AI touched, Shield builds a responsibility-chain record — durable, credentialed, and recorded as the work is done, not reconstructed afterward:
Who
A named licensed professional, with their credentials and license (discipline, jurisdiction, number) as held at the moment of signing.
What
Which AI touched which deliverable, and the role it played.
How they stayed in charge
That the professional directed and reviewed the AI-assisted work before it entered the stamped deliverable, not merely reviewed it after.
Recorded to last
Immutable, time-stamped, retained.
It is evidence of accountability, not a judgment of the engineering. The professional’s judgment stays theirs; the record makes it possible to show they exercised it.
Standards we track
Responsible-charge evidence has to speak the language of the standards that govern licensed practice and the risk behind it. Shield’s methodology maps to each, and we review them quarterly.
AI · Management system
ISO 42001
First international certifiable AI management standard
AI · Risk framework
NIST AI RMF
1.0 + GenAI Profile; 1.0 under revision
EU · Enforcement
EU AI Act
High-risk (Annex III) now Dec 2027 · up to €35M / 7%
DoD · Cybersecurity
CMMC 2.0
Phase 1 live; Phase 2 suspended Jul 2026
BIM · Information mgmt
ISO 19650
Parts 1–3 in revision (DIS, 2026)
Governance · Principles
IBM AI Principles
Published governance principles
US state · AI
Colorado AI Act
Repealed & replaced by SB 26-189 (eff. Jan 2027); affirmative defense removed
Legal · Precedent
ABA Opinion 512
Jul 29, 2024 · generative AI in professional practice
Standards reviewed quarterly
The positioning
Ask a general-purpose AI whether your work is defensible and you get an optimistic answer and a chat transcript. What a board or a carrier wants is different:
| The question | A chat transcript | A responsibility-chain record |
|---|---|---|
| Who was accountable? | Nobody named | A named licensed professional, credentials frozen at signing |
| What did the professional actually do — direct the work, or review it after? | Unknowable | Recorded — direction before, review during, adoption at seal |
| Which license, in which state? | Absent | The specific license held for that project’s jurisdiction |
| Will it hold up years later? | No memory | Immutable, time-stamped, retained |
Who this is for
For the licensed professionals — licensed engineers, architects, and surveyors — who put their seal on work that AI helped produce, and for the firms that carry the professional-liability exposure behind it. If a licensed professional is accountable for AI-assisted work, the record makes that accountability something they can show.
Authorship
The methodology behind Shield is authored and defended by a practicing structural engineer holding PhD, PE, PMP, and IWE credentials — someone who carries the same responsible-charge exposure it’s built to document. It’s engineering rigor applied to a liability question, not a vendor’s marketing claim.
Independent research, published in full and source-graded — honest about what the evidence does not establish.
About the author and the research →The methodology is published. The research is free to read. The tooling is being built in conversation with the profession — with licensed practitioners, and with the carriers and brokers who price this risk. If that’s your world, I’d value the conversation.
Register interest →