Anthropic, the US artificial intelligence company that develops the Claude family of AI models, has officially filed its Form S-1 registration statement with the US Securities and Exchange Commission (SEC), paving the way for a potential stock market listing as soon as mid-November.
The planned initial public offering (IPO) has brought renewed attention to the financial and insurance risks associated with AI, including whether insurers have sufficient capacity to cover potentially significant liabilities arising from the technology.
Commenting on the implications for the insurance sector, Marcos Alvarez, Head of Global Insurance at Morningstar DBRS, said: “Anthropic’s IPO puts the AI insurance protection gap in focus.
“The Company warns in its IPO filing that catastrophic AI liabilities could exceed the insurance industry’s capacity and that coverage for certain risks may be unavailable. From a credit perspective, uninsured liabilities remain a potential claim on the developer’s capital and liquidity.”
Alvarez identified the potential accumulation of claims across policyholders using the same AI models as a key credit consideration for insurers. A widespread model failure or cyber incident could affect multiple customers simultaneously, generating claims across different policies while potentially interrupting insurers’ own operations.
According to the agency, assessing these risks will require insurers to consider their total exposure, policy limits, reinsurance arrangements and ability to absorb losses. Diversification across policyholders may be less effective when customers rely on the same underlying technology, Alvarez noted. The European Insurance and Occupational Pensions Authority (EIOPA) has similarly raised concerns about concentration risks linked to dependence on a small number of AI providers.
Despite the potential for AI to create new insurance opportunities, Alvarez said uncertainty surrounding liability remains a challenge for underwriters. Anthropic’s IPO filing points to unresolved liability rules for autonomous AI agents, as well as uncapped indemnities under certain agreements with external model testers.
The agency also highlighted the possibility that AI-related claims could fall under existing liability policies where specific AI cover has not been purchased. Known as “silent AI exposure”, this could leave insurers facing claims that were not explicitly anticipated when policies were priced. Morningstar DBRS said unclear coverage terms and limited claims data could contribute to inaccurate pricing, uncertainty over reserves and fluctuations in earnings.
Investment performance is another area of concern for the insurance sector. Alvarez said underperformance in AI-related investments could affect insurers through falling asset valuations and liquidity pressures, rather than defaults alone. However, the agency noted that such losses would not necessarily lead to credit rating pressure. The outcome would depend on the scale of insurers’ exposures, the seniority of their investments, available collateral and the strength of their capital and liquidity buffers.
Anthropic’s proposed listing also highlights potential implications for directors’ and officers’ (D&O) insurance. According to Alvarez, the company expects the additional requirements associated with public company status to increase D&O insurance costs and could face reduced policy limits or coverage. Insurers assessing this business will need to consider the quality of disclosures relating to AI capabilities, safety measures and commercial prospects. Morningstar DBRS added that higher premiums should be evaluated alongside the potential for litigation.
AI could nevertheless strengthen insurers’ credit profiles if it delivers lasting efficiency improvements and is implemented with appropriate controls, the agency said. Greater use of AI through partnerships with technology providers could reduce operating costs and improve service quality. However, the financial benefits would depend on whether the savings outweigh the costs of implementation, oversight and correcting errors.
Morningstar DBRS emphasised that insurers would remain accountable for the AI systems they deploy. Effective human oversight, the ability to audit AI-driven decisions and reliable fallback arrangements will therefore be important in managing operational risks as adoption grows.
Alongside this, Evercore ISI analysts have added that “a more immediate issue for the insurance industry will be from latent claims against companies that deploy the technology that could be lurking in general liability or E&O policies.”
The analysts noted that some firms, such as Verisk, have introduced exclusions to limit future exposure, raising questions about liabilities that may remain covered under policies issued before the changes. Potential claims could fall under errors and omissions (E&O), cyber or general liability cover, depending on the nature of the alleged harm, while a gap may persist between demand for AI protection and insurers’ willingness to assume the risk.
The issue has also been highlighted by Andrew Hill, Global FINEX AI Risk Strategic Leader at Willis, following Anthropic’s warning in its S-1 filing that a catastrophic AI incident could create liabilities exceeding the capacity of any single insurer, or potentially the insurance industry as a whole.
He added: “Anthropic’s comments reinforce a growing reality – AI is a frontier risk capable of producing incidents at a frequency and severity beyond anything previously experienced.
“The fundamentals of insurance as a risk transfer strategy for AI risk remain robust, but as our understanding of the risk deepens, particularly its potential for systemic and aggregated losses, it cannot be assumed that all AI exposures will be insurable.”
