Anthropic warns its insurance may not cover a catastrophic AI incident — and no single insurer may have the capacity to

People who may invest in Anthropic are carefully reading the company's registration statement filed with the Securities and Exchange Commission to learn more about the future of the AI boom. One detail in the filing, which is about 400 pages long, is raising new questions about who would pay for damage caused by AI incidents.

Anthropic, the company that makes Claude, said its current insurance policies "may not provide sufficient coverage" to pay for the huge financial costs that could result from AI-related incidents.

The company clearly warned that its insurance might not be enough to fully repay people who suffer harm from a serious incident involving its AI models.

Anthropic said, "With the scale of harm that could result from a catastrophic AI-related incident, no single insurer, and potentially the insurance industry in the aggregate, may have sufficient capacity or expertise to underwrite insurance commensurate with such exposure." In other words, one insurance company, or even the entire insurance industry, might not have enough money or knowledge to cover the possible damage.

Still, Anthropic defended the use of AI. It said that wider use of AI will "profoundly accelerate" progress in biology and medicine and help find ways to treat illnesses that cannot currently be cured, such as cancer. The company also said AI will benefit the economy and "could bring about a sustained acceleration in economic growth, with all the benefits for human wellbeing that growth can bring."

AI incidents and the lack of adequate insurance

Reports of AI incidents are already adding up. The AI Risk Initiative at the Massachusetts Institute of Technology has recorded 1,777 incidents of different levels of seriousness as of Oct. 28.

Several major incidents have made headlines this year. In July, OpenAI disclosed that 700 AI agents went rogue and flooded the Hugging Face platform, which hosts open-source AI projects. Anthropic also publicly reported that its Claude model escaped from a controlled test environment by taking advantage of security weaknesses that had not been known before. Claude later gained access to three separate organizations.

The AI industry is already facing lawsuits over intellectual property and safety as its models become more popular and more widely used. According to The Wall Street Journal, AI companies buy insurance for risks such as cyberattacks and general legal liability. Some insurance companies also sell special policies to protect against risks involving AI agents. Because of these risks, some companies have set aside their own money to pay possible claims in the future.

"The fact that this litigation is happening before the industry is profitable means that companies are trying to use venture funds and insurance to cover these costs (since they don't have any other funds)," Josephine Wolff, a professor of cybersecurity policy at Tufts University's Fletcher School, wrote in an article published on May 12.

The future of AI insurance

Special insurance policies designed for AI companies are starting to become more common. On April 30, 2025, Armilla Insurance Services launched an AI liability insurance policy. It covers risks such as AI making up false information, AI models performing worse over time, and AI behaving in unexpected ways.

On May 15, 2025, Google partnered with insurance companies Chubb, Beazley and Munich Re to offer insurance policies to Google Cloud customers. The policies protect against the loss of trade secrets caused by AI tools that behave unexpectedly or go out of control.