AI Tools and Your E&O Policy: What Design Professionals Need to Know Right Now

Artificial intelligence tools have become a fixture in architectural and engineering practice. Firms use them to accelerate design exploration, generate specifications, analyze building codes, and produce documentation faster than was possible even a few years ago. The productivity gains are real. So is the emerging liability exposure, and now the insurance market is responding in ways that every design professional should understand before their next policy renewal.

Starting January 1, 2026, major changes to professional liability coverage for architects and engineers are taking effect across the U.S. insurance market. Insurers are introducing generative AI exclusions that could leave design firms without coverage for claims arising from work that incorporates AI-generated content or recommendations. If your firm uses these tools, and the odds are good that it does, this is not a development you can afford to ignore.

The Insurance Market Is Moving Quickly

Verisk, the data analytics company whose ISO policy forms serve as the standard templates for much of the U.S. commercial insurance market, introduced new generative AI exclusion endorsements effective January 2026. These forms give carriers the ability to exclude losses arising from generative AI from commercial general liability policies. According to Verisk, the exclusions came in response to strong demand from insurers seeking underwriting tools to address this emerging risk, and carrier interest in adopting them has been described as strong.

The ISO forms are not limited to general liability. AI exclusion language is also being introduced across Directors & Officers, Errors & Omissions, and professional liability policies. For architects and engineers, E&O coverage is the critical line, and it is precisely where the market is showing the most aggressive movement.

At least one major carrier has introduced what has been described as an “absolute” AI exclusion for E&O policies, with language broad enough to bar coverage for any claim arising from the use, deployment, or development of artificial intelligence, regardless of whether the AI system is owned by the firm, licensed from a third party, or embedded in software the firm uses routinely. Several other carriers, including major surplus lines markets, have filed their own AI exclusion language with state regulators, and a wave of additional filings is anticipated through 2026.

Why AI Creates Specific Liability Problems for Design Professionals

The core liability problem with AI tools is well-documented: generative AI systems produce inaccurate outputs, commonly called hallucinations, at rates that should alarm any professional whose work carries legal accountability. Research from Stanford Law School indicates that general-purpose AI tools produce incorrect or fabricated information at rates between 58% and 88%. Even AI tools marketed specifically to professional services users show error rates in the range of 20% to 33%.

For design professionals, these error rates translate into concrete professional liability exposures. An AI tool might recommend structural approaches that violate applicable building codes. It might generate specifications calling for materials that do not exist or that are no longer manufactured. It might produce load calculations that contain errors invisible until a structure is under construction or, worse, occupied. The professional who seals and stamps the drawings remains legally responsible for their accuracy, regardless of what tools were used to produce them. Licensing boards and courts have consistently held that professional accountability does not transfer to an AI vendor.

The liability concern extends beyond design outputs. AI tools used to review contracts, interpret code requirements, assess constructability, or communicate with clients can also produce errors with professional consequences. Confidentiality is another dimension: submitting proprietary project data or client information to a public AI model may expose a firm to liability if that information is later disclosed. The Lloyd’s Market Association has flagged data protection and confidentiality breaches as a distinct category of AI-related professional liability exposure for design firms.

Adoption Is Outpacing Risk Awareness

The timing of these insurance market changes is significant because AI adoption among design professionals has accelerated substantially. Research from the American Institute of Architects published in March 2025 found that approximately one-third of architecture firms had incorporated AI tools into their daily workflows. In a survey conducted by the Royal Institute of British Architects, AI adoption among architects stood at approximately 41%, with nearly half of respondents indicating they believed AI could assist in the design of complex building projects.

Firms are using AI at the early design stage to generate conceptual renderings, in parametric and generative design applications, for routine modeling tasks, and for administrative functions including specification writing and document preparation. The speed and scale of adoption mean that many firms are already relying on these tools before they have assessed what that reliance means for their professional liability coverage.

This gap between operational practice and risk awareness is exactly the situation that tends to produce coverage disputes. If a claim arises from AI-assisted work at a firm that never reviewed its policy for AI-related exclusions, the discovery that the claim is not covered will come at the worst possible moment.

What Design Firms Should Do Before the Next Renewal

The most immediate step for any architecture or engineering firm is a thorough review of current policy language before the next renewal. Policies renewing in the first half of 2026 are the most immediately at risk of encountering new AI exclusion language, but the market shift is broad enough that no renewal cycle is safe to assume unchanged terms.

When reviewing policy documents, search specifically for language referencing artificial intelligence, machine learning, algorithms, automated decision-making, generative AI, and neural networks. These terms may appear in exclusions, conditions, or definitions sections. Policies that are currently silent on AI are not necessarily safe: some carriers are treating silence as an opportunity to add exclusionary endorsements at renewal without prominent disclosure.

Firms should also document their AI governance practices. This means having clear internal policies about which AI tools are approved for use, which project tasks they may be applied to, and what human review and verification processes are required before AI-generated content is incorporated into professional deliverables. Some carriers are likely to take into account the existence of AI governance frameworks when underwriting design professional accounts, and documented oversight practices may become a prerequisite for maintaining full coverage as the market evolves.

Contract language is a related consideration. Design firms that use AI tools in client work may need to address that use explicitly in their professional services agreements, including provisions clarifying how AI outputs are verified and who bears responsibility if an AI-generated element of the work is found to be incorrect. Early disclosure to clients and clearly defined verification responsibilities can reduce the risk that an AI-related error leads to a professional liability claim in the first place.

The Parallel to Silent Cyber Coverage

Industry analysts have drawn an instructive comparison between what is happening with AI exclusions now and what happened with cyber coverage between roughly 2015 and 2023. During that period, insurers discovered they were inadvertently covering cyber losses under property and casualty policies that were never designed to address those risks. The response was systematic: explicit cyber exclusions were introduced, and a standalone cyber insurance market developed to fill the resulting coverage gap.

The AI exclusion movement appears to be following a similar trajectory. General liability and E&O policies were written before generative AI existed as a practical tool, and their language does not clearly address it. Insurers are now moving to make coverage intent explicit, and the direction is exclusionary. A standalone AI liability market is beginning to emerge in response, though coverage scope, pricing structures, and underwriting requirements remain variable and in development.

For design professionals, the lesson from the cyber analogy is that the window for proactive action is limited. Firms that reviewed their cyber exposure and obtained appropriate coverage before markets hardened were better positioned than those that addressed it only after a loss. The same principle applies here.

The Bottom Line for Design Professionals

The professional liability insurance landscape for architects and engineers is shifting in direct response to the AI tools now embedded in everyday practice. The exclusions taking effect in 2026 are not theoretical; they represent a material change in what your policy may cover when a claim arises from AI-assisted work.

Design firms that are proactive about understanding their current coverage, documenting their AI practices, and working with experienced professional liability specialists to address any gaps will be far better positioned than those who discover the problem after the fact. Give us a call to discuss your situation and make sure your coverage reflects the way your firm is actually working.