AI agent certification startup AIUC raises $40M to start auditing frontier models

AIUC, the AI agent certification startup formally known as Artificial Intelligence Underwriting Company, announced on Tuesday that it has raised $40 million in new funding to begin auditing frontier AI models. Until now, its certification work covered only the agents that companies build on top of those models.

The standard at the center of the business is called AIUC-1. Certification means running an agent through roughly 5,000 combinations of risk and attack tailored to the type of business deploying it, with jailbreaks, hallucinations, and data leaks among the failures tested for. Each agent is audited independently and recertified every quarter as attack techniques change, and most of the testing is done by AI agents with humans verifying the final audit, according to TechCrunch.

AIUC argues that uncertainty over risk now slows AI adoption more than any shortfall in what the technology can do. Co-founder and CEO Rune Kvist said most enterprises have a list of agents that were approved in pilots but stalled at the security review, and that proof of security and reliability has become the main bottleneck.

The customer list shows the certification is already being adopted. Voice AI company Eleven Labs used it in February to secure what it described as first-of-its-kind insurance for AI agents, including coverage for a voice agent that gives a customer incorrect information. Anysphere, the developer of Cursor, and legal AI company Harvey AI hold the certification, as do KPMG, Lovable Labs, UiPath, and customer service software maker Fin, formerly Intercom. A consortium of more than 250 security and risk leaders from Fortune 1000 companies shapes the standard and pushes for its adoption inside their own organizations.

The founders came to the problem from different corners. Kvist was Anthropic's first product hire, while co-founder Rajiv Dattani spent time as a partner in McKinsey's insurance practice and then served as chief operating officer of the AI model evaluation nonprofit METR. Dattani compared the effort to the insurance industry's response to early electrical fires: the insurers paying the bill funded Underwriters Laboratories to test and certify products, and the UL mark is still on most light bulbs in America. AI needs the same combination of standards, testing, and insurance, he said.

Ribbit Capital led the Series A with participation from First Harmonic. Nat Friedman at NFDG led a $15 million seed round when the company launched in July 2025, bringing total investment in AIUC to $55 million. Micky Malka, founder and managing partner of Ribbit Capital, said his firm has spent more than a decade backing financial services companies in a sector where trust counts more than anything else, and that AI is heading down the same road faster than the systems companies use to evaluate it.

AIUC started with applications because security reviews blocked rollouts there first, but it says the same pressure is now building around frontier models. The proceeds from the round will extend its audit and insurance work to that layer as part of what AIUC describes as an ecosystem for frontier oversight of the most advanced AI systems.

What matters here is that AI safety is being productized, turning from a voluntary pledge into a third-party certificate that companies can buy, renew, and insure. If quarterly recertification and insurance terms become a precondition for enterprise agent purchases, control over the auditing standard could influence which AI products reach corporate buyers more than any model leaderboard. Watch for whether the frontier-model methodology is published, whether insurance can cover failures at the model layer, and how this privately run certification regime connects to the regulatory frameworks governments are still debating.