The prospect of major AI companies entering public markets is putting fresh attention on the directors and officers (D&O) risks tied to their business claims, disclosures and readiness for life as public companies.
Anthropic is preparing for an IPO, while OpenAI has said it will not go public in 2026, citing AI safety concerns. The potential listings come as investors scrutinize whether AI spending is generating promised revenue or cost savings, and as securities litigation remains a concern for newly public companies.
Julio Bermudez (pictured), who leads Relation’s private equity and transactional risk practice, said IPO-related exposure often starts with the offering documents and whether a company has the structure to meet public-market demands.
The most pertinent questions to suss out D&O risk revolve around what a company has told investors about its AI strategy, whether it can substantiate those claims, and whether its insurance program reflect the scale and complexity of the risk.
Bermudez said companies built around an AI thesis could face shareholder scrutiny if the business results do not match what investors were told to expect. The question is whether a company can show that AI is producing the revenue growth or expense reductions underpinning its strategy.
“If it’s achieving neither, does the company then have to take what seems like the easiest path — expense reduction — to justify its thesis on AI? Does it become a self-fulfilling prophecy?” Bermudez said.
He cautioned that it is too early to attribute rising securities litigation activity to AI-related allegations without examining the details. Still, AI spending and its business impact are likely to draw attention as companies make claims about expected productivity gains, new revenue or lower costs.
Brokers are also reviewing policy language around AI-related risks. “From an insurance standpoint, we’re also seeking to make sure there’s affirmative coverage in specific areas,” Bermudez said. “You start talking about cyber in particular and other elements like that, and you want to make sure there aren’t extraneous exclusions that could adversely affect the D&O contracts.”
AI companies face the same operational challenge as other IPO candidates: moving from private-company practices to the reporting, governance and staffing demands of a public listing.
Bermudez recalled a de-SPAC involving a business whose senior executives had been accustomed to operating with a small finance team. The company struggled to adapt to public-company requirements, later faced a securities class action lawsuit and eventually declared bankruptcy. Those experiences underscore why brokers examine the company’s management bench as well as its disclosures. “Whether the organization is sophisticated enough to go public and has the bench strength to manage life as a public company” is a central exposure question.
Brokers are often brought in once the S-1 draft process is underway, though Bermudez said they may receive an earlier heads-up under a nondisclosure agreement. This gives them time to model different coverage structures and compare limits and pricing before approaching insurers.
Options can include a traditional public-company D&O program, a combination of primary coverage and excess Side A, or a Side A-only program for a company viewed as higher risk or carrying a substantial valuation. The right structure depends on the company’s circumstances and what the market will support.
“We want to stay ahead of it and have conversations with senior management and leaders so there are no surprises along the way,” Bermudez said.
Public-company D&O is a competitive, relatively soft market, even as securities class action activity remains elevated and legal costs rise. However, defense costs can climb quickly, particularly when cases move into discovery or survive a motion to dismiss.
Market competition can create room for brokers to seek lower pricing, retentions or broader coverage. But the cost of litigation also raises questions about whether historical benchmarks and limits remain adequate for today’s claims. From their side, brokers are increasingly using peer comparisons, recent market data and modeling of historical class action activity to help clients assess limits. The analysis informs decisions, though it does not produce a single answer for every company.
“Are we pigeonholing ourselves by relying on old data about benchmarking, securities claim costs and activity?” Bermudez said. “Does that data need to be refreshed to reflect a new mindset?”
The pressure also creates a tension for underwriters. “I’m not saying underwriting discipline disappears,” Bermudez said, “but, given the situation, insurers may have to take chances on risks they could walk away from in a different environment.”
