AIUC, a startup that creates insurance for AI agents, emerges from stealth with a $15 million seed

That’s right: AI Agent Insurance Contracts. AIUC co-founder and CEO Rune Kvist says that agent insurance — autonomous AI systems that can make decisions and take action without human supervision at all times, are likely to become a big company. The founding team of KVIST, which previously adopted the first product and commercial employment in humanity in 2022, also includes Thiel Fellow, who previously founded the consumer underwriting business, and former CTO Brandon Wang, Rajiv Dattani. McKinsey The partner who worked in the global insurance sector was COO of Metr, a research nonprofit that evaluated Openai and human models before its deployment.

Creating financial incentives to reduce the risk of AI agents recruiting

At the heart of AIUC’s approach is a new risk and safety framework called AIUC-1, designed specifically for AI agents. We bring together existing standards, including the NIST AI risk management framework, EU AI law, and Miter’s Atlas threat model, including the Atlas, an auditable layer of agent-specific safeguards. The idea is simple. It makes it easy for businesses to employ AI agents with the same type of trust signals expected in cloud security and data privacy.

“The key to insurance is to create financial incentives to reduce risk,” Kvist said. luck. “So we’re tracking and where are you doing wrong, what’s the problem you’re solving, and insurers can force you to take certain steps to get certified.”

While other startups are currently working on AI insurance products, Kvist said there is no one that has built a type of risk-preventing agent standard like AIUC-1. “Insurance and standards are closely related to building trust in AI recruitment,” he said.

“AIUC-1 creates standards for AI adoption,” said John Bautista, a partner at the law firm Orrick who helped create the standard. “As businesses enter a brave new world of AI, there is a lot of legal ambiguity to keep adoption up. With new laws and frameworks constantly emerging, companies need one clear standard that brings it all together and makes adoption very simple,” he said.

The need for independent vendors

The story of American progress is also a story of insurance, he added. Benjamin Franklin has established the country’s first mutual fire insurance company in response to a catastrophic home fire. In the 20th century, professional players like UL Labs emerged from the insurance industry to test safety of electrical appliances. Auto insurance companies have set the crash testing standards that have created the modern automotive industry.

AIUC bets that history is about to repeat itself. “it’s not Toyota It will do car accident tests, it is an independent body. ” Kvist said, “I think we need an independent ecosystem of companies asking questions. Can we trust these AI agents?”

To achieve that, AIUC offers Trifecta: Standards, Audits and Liability Compensation. The AIUC-1 framework creates technical and operational baselines. Independent audits test real performance. They try to fail agents, hallucinate them, leak data, or put them at risk. Insurance contracts cover customers and vendors in cases where agents are harmed, and pricing reflects the safety of the system.

This type of insurance policy could cover fallout if an AI sales agent accidentally publishes personally identifiable information, or if a financial AI assistant creates an insurance contract or misquots tax information. According to Kvist, financial incentives are the key. To ensure that consumers get better car insurance rates for having airbags and anti-lock brakes, AI systems that pass the AIUC-1 audit can get better terms with insurance, in Kvist’s view. This will allow AI vendors to be directed towards better practice, faster, and give businesses specific reasons to adopt them faster before competitors do it.

Use insurance to adjust your incentives

AIUC’s view is that not only governments but markets can promote responsible development. Top-down regulations “it’s difficult to get right,” Kvist said. But we also leave everything to companies like Openai, Anthropic and Google. And voluntary safety commitments have already returned. Insurance will coordinate incentives and create a third way of evolving with technology, he explained.

Kvist compares AIUC-1 to SOC-2. This is a security authentication standard that gave startups a way to signal trust to enterprise buyers. He imagines the world where AI agent liability insurance is common today and is necessary. It is today’s cyber insurance, forecasting a $500 billion market by 2030, and even covering cyber insurance.

AIUC has already worked with customers and insurance partners from multiple companies (AIUC says it can disclose its name).

Investors like Nat Friedman agree. As a former CEO of Github, Friedman saw firsthand the trust issues when launching Github Copilot. “All of his customers were wary of adopting it,” recalls Kvist. “There was all of these IP risks.” As a result, Friedman had been looking for an AI insurance startup for several years. After a 90-minute pitch meeting, he said he wanted to invest. I moved to join Alexandr Wang Mark Zuckerberg’s new Meta Superintelligence Labs.

Kvist said he would ensure that AI agents will become mainstream in a few years. “These agents have a much bigger promise: ‘We’re going to do our job for you,” he said. “We believe that responsibility will be much greater and therefore the interest will be much greater.”