Startup raises $15M for AI liability insurance framework amid industry concerns

A startup has emerged to address a growing concern in the AI industry: liability when autonomous systems malfunction. The Artificial Intelligence Underwriting Company (AIUC) is launching from stealth with a $15 million seed round led by Nat Friedman at NFDG, supported by Emergence, Terrain, and notable angel investors including Anthropic cofounder Ben Mann and former CISOs from Google Cloud and MongoDB [1]. The company aims to establish insurance, audit, and certification infrastructure to safely integrate AI agents into enterprise environments.

AIUC’s cofounder and CEO, Rune Kvist, previously a key hire at Anthropic, emphasizes that autonomous AI systems—capable of decision-making without constant human oversight—require robust risk management. The company’s team includes CTO Brandon Wang, a Thiel Fellow with a consumer underwriting background, and Rajiv Dattani, a former McKinsey partner with expertise in global insurance and AI model evaluation [1]. Their approach centers on a risk framework called AIUC-1, which synthesizes existing standards like NIST’s AI Risk Management Framework, the EU AI Act, and MITRE’s ATLAS threat model while introducing agent-specific safeguards. This framework aims to provide enterprises with the same level of trust as cloud security certifications [1].

The insurance model leverages financial incentives to mitigate risks. By tracking and certifying AI systems’ compliance with safety protocols, insurers can enforce best practices for coverage, Kvist explained to Fortune. This mirrors how car insurers developed crash-test standards that became industry benchmarks. Similarly, AIUC envisions a future where independent bodies—not AI developers—audit agents for risks like data leaks, hallucinations, or dangerous actions [1].

John Bautista, a partner at Orrick who contributed to AIUC-1’s development, noted the framework simplifies adoption by consolidating legal ambiguities into a single standard. “As businesses navigate AI’s complexities, they need a clear benchmark to act,” he said [1]. AIUC’s trifecta of standards, audits, and liability coverage allows enterprises to assess AI agents’ safety through real-world stress tests before deployment. For example, an AI sales agent exposing customer data or a financial AI misquoting tax information could trigger insurance claims. Insurers would adjust premiums based on audit results, incentivizing developers to prioritize safety [1].

The startup positions itself as a market-driven solution to AI governance, contrasting with top-down regulation or voluntary commitments. Kvist likens AIUC-1 to SOC-2, a security standard that boosted trust in software startups. He predicts AI agent liability insurance could become as ubiquitous as cyber insurance, potentially reaching a $500 billion market by 2030 [1].

Investor support underscores the potential. Nat Friedman, former GitHub CEO, recognized the need for AI insurance during the launch of GitHub Copilot, where intellectual property risks stoked client concerns. After a 90-minute pitch, Friedman led the seed round, citing AIUC’s alignment with Meta’s recent Superintelligence Labs initiative [1]. The company is already collaborating with enterprise clients and insurers, though it has not disclosed names.

Kvist envisions a future where insuring AI agents becomes mainstream, driven by their increasing integration into critical workflows. “The liability potential is significant, and so is the market interest,” he said [1]. By creating a third-party ecosystem of trust, AIUC aims to accelerate enterprise adoption of AI agents while addressing risks through a blend of standards, audits, and financial accountability.

Source: [1] [title1Exclusive: Who covers the damage when an AI agent goes rogue? This startup has an insurance policy for that] [url1https://fortune.com/2025/07/23/ai-agent-insurance-startup-aiuc-stealth-15-million-seed-nat-friedman/]