A startup founded by an early Anthropic employee and the former chief operating officer of AI safety research group METR has raised $40 million to build a third-party certification system for AI agents. The Artificial Intelligence Underwriting Company, known as AIUC, announced the Series A on Tuesday, led by Ribbit Capital with participation from First Harmonic, according to TechCrunch. Combined with an earlier $15 million seed round, the company has now raised $55 million.

AIUC was founded by Rune Kvist, a former Anthropic employee, and Rajiv Dattani, who served as METR’s COO from 2024 to 2025 and remains on its board. The two are brothers-in-law. Their pitch to investors rests on a specific failure of the enterprise AI market: capable models are not the bottleneck for adoption anymore, trust is. “Banks, hospitals, governments and militaries no longer decline to deploy AI because a model isn’t smart enough,” Kvist told TechCrunch. “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.”

The company’s answer borrows directly from cybersecurity. Just as SOC 2 became the compliance bar that lets enterprise software vendors prove they handle data responsibly, AIUC built a parallel standard called AIUC-1 for AI agents. To decide what the standard should test, the company says it assembled a consortium of about 250 risk and security leaders, the people who actually sign off on agent purchases inside large organizations.

Vendors that go through AIUC’s process get run through around 5,000 test scenarios probing jailbreaks, hallucinations, and data leakage, producing a report the company says runs about 100 pages. AIUC uses AI agents to execute the tests and AI to analyze the resulting data, though the company says humans verify the final audit before it goes out. AIUC lists Cursor, Lovable, Harvey, and ElevenLabs among its customers, claims made by the company itself and not independently verified.

The timing lines up with a broader shift at the frontier labs. Anthropic CEO Dario Amodei recently argued the AI industry should slow the pace of frontier development, pointing to a rise in agent misbehavior, and raised the idea of obliging labs to embed outside evaluators, with METR named as one possibility. AIUC is not proposing to embed staff at customer sites the way Amodei described. Its model is closer to a one-time or periodic audit than continuous oversight, a narrower promise that is also easier to sell to enterprise buyers who need a certificate today rather than a monitoring relationship.

That gap between audit and oversight is where AIUC’s real test will happen. A 100-page report generated largely by AI tools reviewing AI agents asks buyers to trust a process that has not existed long enough to have a track record of catching failures before they happened in production. Dattani frames the value as informed risk, not a guarantee: “Here’s where it passes and where you can trust it. And here’s where there’s concerns,” he told TechCrunch. “You should be aware of those references before you make the decision to buy.”

For enterprise buyers currently evaluating coding agents or customer-facing AI tools, an AIUC-1 certificate is now a data point worth requesting alongside a vendor’s own benchmark claims, though it is not yet the industry-wide standard SOC 2 became over a decade of adoption.

Reported by Julie Bort for TechCrunch, September 15, 2026.