Over the last several years, artificial intelligence (AI) has evolved into a central component of many companies’ growth strategies. As organizations increasingly integrate AI into their operations, products, and business models, the associated litigation risks have begun to emerge as well. The D&O Diary has been tracking the rise of AI-related litigation, from early AI-washing cases to a growing number of securities suits involving AI infrastructure investments, AI-enabled business models, and AI-related disclosure issues.

The emergence of these lawsuits raises an important question for D&O insurers and underwriters. How should they evaluate AI-related risk? More specifically, can underwriters realistically keep pace with a technology that is evolving as rapidly and broadly as artificial intelligence? Thus, AI presents challenges that may not lend themselves to traditional underwriting approaches. Indeed, the more interesting question may be whether AI represents a category of risk that can be understood and assessed through conventional underwriting methods.

The Litigation Has Moved Beyond AI-Washing

When AI-related litigation first began attracting attention, many securities claims involved what became known as “AI-washing,” allegations that companies exaggerated their AI capabilities or overstated the role artificial intelligence played in their businesses.

However, the plaintiffs’ bar has already moved beyond traditional AI-washing theories. Securities lawsuits have been filed against AI infrastructure companies, data center operators, power generation providers, lending platforms, and software companies whose AI-related initiatives allegedly failed to perform as represented. In these cases, AI is often not the central issue. Instead, AI becomes intertwined with disclosures concerning revenue growth, capital expenditures, operational efficiency, customer adoption, or business strategy.

That distinction matters. These cases increasingly resemble traditional securities lawsuits involving disclosure issues and strategic execution rather than technology-related claims. The allegations often center on whether management accurately described the opportunities and risks associated with AI initiatives and whether investors received sufficient information regarding the challenges involved in implementing those strategies.

As AI becomes more deeply embedded in corporate operations, it seems likely that this trend will continue. Future litigation may increasingly focus on governance, oversight, capital allocation, disclosure practices, and business judgment rather than on the technology itself.

AI Is Not One Risk

One of the challenges facing underwriters is that artificial intelligence is often discussed as though it were a single category of risk. However, AI encompasses a wide range of technologies with significantly different risk profiles.

Many organizations today deploy predictive AI, generative AI, agentic AI, autonomous systems, or combinations of several different technologies. While these systems are often grouped together under the broad heading of artificial intelligence, they create distinct operational, regulatory, and governance concerns.

Predictive AI has been used for years in industries such as healthcare, financial services, and insurance. These systems rely on historical data to identify patterns and forecast outcomes. The associated risks often involve bias, inaccurate predictions, model drift, and regulatory scrutiny regarding decision-making processes.

Generative AI creates an entirely different set of concerns. Because these systems generate content, they raise issues involving intellectual property rights, misinformation, hallucinations, copyright infringement, and disclosure accuracy. Many of the legal disputes currently attracting attention involve these types of risks.

Agentic AI introduces yet another layer of complexity. These systems increasingly are designed not merely to generate information but to take actions, make decisions, and interact with other systems with limited human intervention. Questions regarding accountability, oversight, authorization, and control become increasingly important as organizations deploy technologies capable of acting independently.

Autonomous systems present perhaps the broadest set of challenges. Whether operating in transportation, manufacturing, logistics, healthcare, or critical infrastructure, autonomous technologies can create consequences that extend far beyond traditional technology-related losses and implicate multiple lines of insurance.

The diversity of these technologies helps explain why AI underwriting remains a work in progress. Underwriters are not evaluating a single risk. They are evaluating an evolving collection of risks that continue to change as the technology itself develops.

What Underwriters Can Underwrite

Many AI-related D&O exposures involve areas that underwriters have evaluated for years. The key questions frequently involve governance, oversight, disclosure controls, and risk management rather than the underlying technology itself.

For example, underwriters increasingly may focus on how central AI is to a company’s business model. Is AI a core driver of revenue growth? Is it embedded within existing products? Or is it still experimental? Companies that position AI as a primary growth engine may face different disclosure risks than organizations using AI as a supporting operational tool.

Similarly, governance practices may become increasingly important. Does the board receive regular updates regarding AI initiatives? Are there established governance frameworks governing AI deployment? How are risks escalated? What controls exist around model validation, data governance, vendor oversight, and regulatory compliance?

Disclosure practices may prove especially important. Many of the AI-related lawsuits filed to date involve allegations that companies overstated opportunities, understated risks, or failed to adequately explain the limitations associated with AI initiatives. As a result, underwriters are likely to pay increasing attention to the relationship between a company’s internal understanding of its AI capabilities and its public statements regarding those capabilities.

These are all familiar D&O underwriting concepts. While the technology may be new, the underlying concerns surrounding governance, oversight, and disclosure are not.

The Limits of Underwriting AI

The challenge is that AI risk is not necessarily confined to a company’s own activities. An organization may face AI-related risks arising from competitors, customers, vendors, regulators, or broader market developments that are difficult to predict and even harder to quantify.

A company’s AI strategy may succeed, but a competitor’s strategy may prove more effective. Customers may change their purchasing behavior because of AI-driven market shifts. Regulators may introduce new compliance obligations. Vendors may alter the functionality of third-party AI systems on which companies rely. In some industries, the greatest AI-related risk may not come from adopting AI too aggressively but from failing to adapt quickly enough.

Traditionally, when underwriters encounter uncertain risks, they have several options available. They can avoid exposure, narrow coverage, or charge additional premium to compensate for uncertainty. None of those solutions is particularly easy in the current environment.

AI is becoming so pervasive that avoiding any exposure from AI entirely may be impossible. Even companies that are not actively developing AI products are increasingly affected by AI-driven changes occurring throughout their industries. As a result, underwriters may find themselves confronting a category of risk that is both highly consequential and inherently difficult to model.

This reality suggests that a degree of underwriting humility may be appropriate. Artificial intelligence is not merely an emerging technology risk; it is increasingly reshaping business models, competitive dynamics, customer behavior, regulatory expectations, and capital allocation decisions across industries. Some AI-related risks may be identifiable and measurable, while others may emerge from factors beyond a company’s direct control.

A competitor’s AI strategy, a regulator’s response to evolving technology, or a fundamental shift in customer expectations may ultimately prove as consequential as a company’s own AI deployment. In that environment, the challenge for underwriters may be less about perfectly quantifying AI risk and more about understanding the limits of what can reasonably be known and assessed.

Conclusion

It would be unrealistic to expect a single underwriting framework for AI to emerge in the near future. Artificial intelligence encompasses a broad range of technologies, industries, and use cases. A company developing foundation models presents a very different risk profile than a healthcare provider deploying AI-assisted diagnostics, a lender using predictive analytics, or a manufacturer utilizing autonomous systems. Given those distinctions, it is hardly surprising that insurers are approaching AI-related exposures from different perspectives.

Indeed, the diversity of underwriting approaches may ultimately prove to be a strength rather than a weakness of the D&O marketplace. One of the defining characteristics of the D&O insurance industry has always been the willingness of different insurers to reach different conclusions about the same risk. Differing underwriting philosophies, risk appetites, and pricing models create competition and allow insureds access to a broader range of coverage solutions.

The insurance industry has navigated emerging exposures before, including cyber liability, ESG-related risks, cryptocurrency, and SPAC litigation. In each instance, underwriting approaches evolved as claims experience developed and risk characteristics became better understood. Artificial intelligence may ultimately follow a similar path, though perhaps on a larger scale given its breadth and potential impact across industries.

Over time, AI may give rise to increasingly specialized underwriting approaches. Some insurers may develop expertise in companies that build AI technologies, while others may focus on organizations that use AI as a business tool rather than as a core product. Certain underwriters may place greater emphasis on governance structures and disclosure controls, while others may focus on regulatory exposure, intellectual property concerns, or operational risks associated with AI deployment. As claims experience develops and litigation trends become clearer, underwriting approaches are likely to evolve alongside the technology itself.

The challenge for underwriters is not necessarily to develop a single model for evaluating AI risk. Rather, it may be to determine which aspects of AI-related exposure can be meaningfully assessed while recognizing the limits of what can presently be known. In that respect, the future of AI underwriting may tell us as much about the adaptability of the D&O insurance industry as it does about artificial intelligence itself.

One final note. There is an entirely separate way in which AI represents a challenge to the D&O insurance industry, and indeed to the insurance industry as a whole. That is, corporate leadership must and undoubtedly are scrambling to figure out how to incorporate AI models, tools, and agents into brokering, underwriting processes and claims management. This issue is a huge topic and beyond the scope of this blog post. An issue for another day.