Startups & Funding
Startup AIUC raises $40M to certify AI agents like a SOC 2 for safety
AIUC, founded by an early Anthropic employee and a former METR COO, raised a $40 million Series A to run enterprise AI agents through a safety-certification standard modeled on the cybersecurity audit SOC 2.
A day after Anthropic researcher Jacob Coxon quit his job over concerns that AI could pose catastrophic risk, founders and brothers-in-law Rune Kvist and Rajiv Dattani described a narrower problem they think they can solve: keeping AI agents from going rogue inside enterprises. “AI is getting smarter at an increasingly rapid rate. The surprising thing about AI is that it becomes harder to adopt and harder to control as AI gets smarter, not easier,” said Kvist, an early Anthropic employee. Dattani is the former COO of the AI safety research organization METR, serving in that role from 2024 to 2025 and remaining a board member.
The pair’s startup, Artificial Intelligence Underwriting Company (AIUC), aims to bring AI safety assessment to enterprises and to companies building AI models and agents; it names Cursor, Lovable, Harvey, and ElevenLabs as customers. On Tuesday, AIUC announced a $40 million Series A led by Ribbit Capital, with participation from First Harmonic, adding to a previous $15 million seed round from Nat Friedman’s NFDG along with Emergence, Terrain, and Anthropic co-founder Ben Mann — bringing total funding to $55 million.
AIUC applies a cybersecurity-style audit model to AI risk, using the widely adopted SOC 2 standard as its template. The company built a standard called AIUC-1 and a testing service to validate agents against it, developed with input from a consortium of roughly 250 security and risk leaders who buy agents. “Banks, hospitals, governments and militaries no longer decline to deploy AI because a model isn’t smart enough,” Kvist said. “They decline because they’ve made commitments to their own customers about what a system will and won’t do, and nobody can currently guarantee that.” AIUC runs an agent through a suite of some 5,000 tests covering jailbreaks, hallucinations, and data leaks, producing a roughly 100-page report; AI agents run the tests and analyze the data, but humans verify the final audit.
The approach echoes METR’s own work evaluating frontier labs’ models, though METR’s focus has mostly been on performance rather than safety. Anthropic CEO Dario Amodei has recently called for the industry to pace frontier development and floated requiring labs to use embedded third-party evaluators, naming METR as one possibility. AIUC isn’t proposing to embed itself at customer sites, but Dattani said the goal is similar: giving enterprises an independent read on how safe their AI agents are before they buy.
Why it matters
A third-party certification layer for AI agents could become a prerequisite for regulated industries — banks, hospitals, governments — to adopt AI at all, turning AI safety assessment into its own enterprise-sales category.