Today, the Artificial Intelligence subscription Society (AIUC) emerges from stealth with a $ 15 million seed tower led by Nat Friedman in NFDG, with the participation of Emergence, Land and notable angels, including the anthropogenic co -founder Ben Mann and the former Ciso of Google Cloud and Mongodb. The objective of the company? Build the insurance, audit and certification infrastructure necessary to safely introduce AI agents into the business world.
It’s true: insurance policies for AI agents. The co -founder of AIUC and CEO Rune Kvist says that insurance for agents – that is to say autonomous AI systems capable of making decisions and acting without constant human supervision – is about to be large companies. Previously, the first product and marketing rental at Anthropic in 2022, the founding team of Kvist also includes the CTO Brandon Wang, a Thiel scholarship holder who previously founded a consumer subscription company, and Rajiv Dattani, a former McKinsey partner who assessed Openai and the insurance models, and was the COO of METR, evaluated Openai and anthropist models before work.
Create financial incentives to reduce the risk of adoption of AI agents
At the heart of the AIUC approach is a new risk and security framework called AIUC-1, designed specifically for AI agents. It brings together existing standards such as the risk management framework for AI NIST, the EU AI law and the threat model of the Mitre Atlas – then the layers on verifiable guarantees and specific to the agent. The idea is simple: facilitate companies to adopt AI agents with the same type of trust signals they expect in cloud security or data confidentiality.
“The important thing about insurance is that it creates financial incentives to reduce risk,” said Kvist Fortune. “This means that we are going to follow, where it goes wrong, what are the problems you solve. And insurers can often apply that you take certain measures to be certified.”
Although other startups are also currently working on AI insurance products, Kvist has said that no one is building the type of agent standard that prevents risks like AIUC-1. “Insurance and standards go hand in to create confidence in AI adoption,” he said.
“AIUC-1 creates a standard for the adoption of AI,” said John Bautista, a partner in the ORRICK law firm who helped create the standard. “While companies enter a new courageous AI world, there is a ton of legal ambiguities that maintain adoption. With new laws and managers by constantly emerging, companies need a clear standard that brings together all together and makes adoption massively simple,” he said.
A need for independent sellers
The history of American progress, he added, is also a story of insurance. Benjamin Franklin founded the country’s first party insurance company in response to devastating house fires. In the 20th century, specialized actors such as Ul Labs emerged from the insurance industry to test security electrical devices. Motor insurers have built accident test standards that have given birth to the modern automotive industry.
AIUC bets that history is about to repeat. “It is not Toyota who does the car crash tests, these are independent bodies.” Kvist has underlined. “I think there is a need for an independent ecosystem of companies that answer [the question]Can we trust these AI agents? »»
To get there, AIUC will offer a trifecta: standards, audits and liability coverage. The AIUC-1 frame creates a technical and operational reference. Independent audits are testing real world performance – trying to bring agents to fail, hallucinate, flee data or act dangerously. And insurance policies cover customers and suppliers in the event that an agent causes damage, with prices that reflect system security.
If an AI sales agent accidentally exposes personally identifiable information, for example, or if an AI assistant in finance manufactures a police force or cite tax information, this type of insurance policy could cover the benefits. Financial incitement, said Kvist, is the point. Like consumers obtain a better automobile insurance rate to have airbags and anti-change brakes, AI systems that transmit the AUC-1 audit could obtain better conditions on insurance, according to Kvist. This pushes AI sellers to best practices, more quickly – and gives businesses a concrete reason to adopt earlier, before their competitors.
Use insurance to align incentives
AIUC’s opinion is that the market, not just the government, can stimulate responsible development. Descending regulations are “difficult to do well,” said Kvist. But leaving all of this to companies like Openai, Anthropic and Google do not work either – volument safety commitments are already back. Insurance creates a third way to align incentives and evolves with technology, he said.
Kvist compares AIUC-1 to SOC-2, the safety certification standard that has given startups a way to report confidence in business buyers. He imagines a world in which the civil liability insurance of AI agents becomes as common – and necessary – as cyber -assurance is today, predicting a market of $ 500 billion by 2030, even eclipses cyber -assurance.
AIUC is already working with several business customers and insurance partners (AIUC said that it could disclose the names) and moves quickly to become the reference of the industry for the safety of AI agents.
Investors like Nat Friedman agree. As a former CEO of GitHub, Friedman saw the confidence emitting first -hand during the launch of Github Copilot. “All of his customers were wary of adopting it,” recalls Kvist. “There were all these risks of intellectual property.” As a result, Friedman has been looking for an AI insurance startup for a few years. After a 90 -minute meeting, he said that he wanted to invest – what he had done, in a seed tour in June, before Friedman moved to join Alexandr Wang in the new meta -superintelligence laboratories of Mark Zuckerberg.
In a few years, said Kvist, assuring that AI agents will be dominant. “These agents make a much larger promise, which is” we are going to do the work for you, “he said. “We believe that responsibility becomes much greater, and therefore interest is much more important.”
