Anthropic, the developer behind the Claude artificial intelligence ecosystem deeply embedded across Asia-Pacific financial, legal, healthcare, and insurance sectors, submitted its initial public offering (IPO) prospectus this week with stark warnings about the technology’s broader implications. In its filing for a Nasdaq listing—projected to surpass SpaceX’s US$1.8 trillion valuation—the firm stated that AI technology could present “catastrophic or existential risks to humanity” if allowed to develop without adequate safeguards. The prospectus disclosed first-half losses of US$80.7 billion against revenues of US$16.3 billion, characterising its corporate strategy as “bold but sane.”
Crucially, the regulatory submission serves as a binding legal document that explicitly admits liability uncertainty. Anthropic acknowledged that it remains unsure about where and how it might be held legally responsible for actions performed autonomously by its systems. The company listed regulatory interventions stemming from AI outputs—including critical issues such as child safety and suicide—as primary risk factors, alongside an emerging wave of cybersecurity incidents linked to autonomous AI agents.
This admission poses an immediate dilemma for insurance underwriters across Asia-Pacific hubs, including Singapore, Hong Kong, Japan, Australia, and India, where Claude has been integrated into core business operations. Insurers now face the reality that the technology creator itself cannot define where liability ultimately rests.
Anthropic categorised three primary exposure types in its filing:
Research from broker Willis highlights that the professional indemnity market is rapidly pivoting from silent AI coverage to explicit affirmative warranties or total exclusions. Alarmingly, one in five insurance professionals reports that their clients have already suffered AI-related losses.
Regulatory bodies across the region are rushing to establish governance frameworks as adoption outpaces oversight. On 7 October, the Monetary Authority of Singapore (MAS) issued AI risk management guidelines covering banks, insurers, intermediaries, and payment platforms. In India, the Insurance Regulatory and Development Authority (IRDAI) formed a seven-member working group in June 2026 to craft the country’s first formal AI governance structure, while the Hong Kong Monetary Authority (HKMA) and Japan’s Financial Services Agency (FSA) advance their own supervisory mechanisms.
With OpenAI expected to launch its own IPO early next year, insurers offering D&O, technology errors and omissions (E&O), and cyber coverage are under urgent pressure to re-evaluate whether legacy policy language remains fit for purpose.
