Most conversations about AI liability in professional services generate more alarm than insight. For architects and engineers, the starting point is more grounded - and more protective - than the general conversation suggests.
"Artificial intelligence right now is really more of a tool toward delivering instruments of service (plans and specifications)," said Kevin Collins, A&E practice leader at Victor . "As an architect or engineer, you're legally required to sign and seal the documents to confirm that what you've designed is constructible and meets the standard of care within that jurisdiction for the project.”
The professional seal is not a formality. It is the accountability mechanism at the center of A&E professional liability, and it does not yield to technology. A firm that uses an AI design tool and then signs and seals the drawings has retained full liability for what those drawings say. The AI is an instrument. The professional is responsible.
"If people are utilizing AI technology in order to design the facility but there’s a design error and they just assume that the AI bot is correct and they sign and seal the drawing, they still have liability for costs, losses, or damages caused by any errors or omissions," Collins said. "But it doesn't change the coverage because the AI resource is an instrument in order to be more efficient, to be more effective, create solutions."
For brokers placing A&E professional liability, the coverage picture on AI is clearer than in many lines.
AI-assisted design work generally falls within existing E&O coverage because the underlying professional service - design judgment, standard-of-care compliance, professional review - has not changed. The AI tool may be treated similarly to the CAD software, structural modeling platforms, and engineering databases that have always been part of professional practice.
"From our perspective, it's covered generally in the policy," Collins said. That clarity flows directly from the human accountability that the profession's regulatory structure requires.
The cyber picture is distinct. AI is driving greater frequency and sophistication of attacks while simultaneously enabling faster threat detection - a different exposure profile that belongs in a dedicated cyber liability conversation rather than the professional liability one.
The cases where AI creates genuine new liability are not necessarily about AI replacing professional judgment. They are about governance failures - situations where an AI-generated output was not reviewed adequately before it became a deliverable.
Collins described a claim example from a recent conference that illustrates the point precisely. A firm used AI during the design process to assist in selecting artwork for a building's foyer. The client approved the design. What the firm did not catch was that the AI had surfaced an image of an existing artist's work, which they then recreated. The claim was settled with the original artist.
"They didn't have internal practices in that communication side of it," Collins said. The error was not an AI error. It was a governance failure - the absence of a review process that would have caught a problematic output before it left the firm.
That case is representative of where A&E firms' AI exposure concentrates: intellectual property risk from unchecked outputs, contractual and reputational risk from disclosure obligations not met, and quality control failures where AI-assisted work is not reviewed to the standard the professional's seal demands.
Victor's risk management guidance for A&E firms organizes the AI response around two distinct areas.
The first is internal governance. Every firm using AI tools needs clear policies on which tools are approved, what data they can and cannot use, and what the review process is before any AI-assisted output is signed, sealed, or delivered. The dataset question is specific. "A lot of it's built around datasets to make sure that it's contained, that you're not using public sources," Collins said. The artwork case illustrates why firms need internal controls to identify potential intellectual property issues in AI-generated outputs, including content derived from publicly accessible sources that may still be protected by copyright or other intellectual property rights.
The second is external communication. Contracts for A&E services are increasingly including AI-specific language - some clients prohibiting AI use outright, others requiring disclosure, others imposing limits on particular applications. "What we're telling our clients is that they're proactive in disclosure to the client," Collins said. "Because a lot of the systems that you use for design have an AI component to them, even if it's just assist for efficiency."
Proactive disclosure is not just a contractual compliance measure. A firm that tells its clients which AI tools it uses, what governance is in place, and how outputs are reviewed before being sealed is managing a client relationship on the front foot. Firms that fail to address applicable disclosure requirements may face contractual disputes or reputational risks if clients subsequently discover undisclosed AI use.
Collins flagged the pace of development as the variable that makes complacency the real risk. "Where we were six months ago internally with AI use versus now is dramatically different. And that's going to continue to go,” he said.
That trajectory means the governance conversation is not a one-time exercise. Policies that were adequate last year may not reflect what the tools are doing now. The firms staying ahead of that curve - reviewing their AI governance regularly, updating their disclosure practices as tools evolve, investing in the quality control processes that make the professional seal meaningful - are not just managing a liability. They are differentiating themselves in a market where clients are increasingly asking which firms have their AI use under control.
Brokers who can walk A&E clients through the governance questions - which tools, what datasets, what review process, what contract language - are providing something that goes well beyond coverage placement. That conversation is the one that builds the long-term client relationship, and it starts with understanding the continuing importance of professional judgment, accountability, and the applicable standard of care.
