AIUC, a startup founded by Rune Kvist and Rajiv Dattani, is building an enterprise AI safety infrastructure through third-party auditing and certification of AI agents, aiming to standardize testing, auditing, and insurance much like UL certification for electrical products.

The founders argue that the main bottleneck for enterprise AI adoption is measuring security and reliability, and that AI safety requires standardized testing and independent auditing with insurance options.

In its Series A, AIUC plans to expand its audits, safety standards, and insurance capabilities for frontier models while continuing to provide independent evidence of AI behavior before large-scale deployment.

Customers such as Cursor, Lovable, Harvey, ElevenLabs, among others, are already using AIUC’s framework.

The funding signals a growing need for coordinated industry standards among developers, enterprises, auditors, and insurers to enable widespread AI risk infrastructure adoption.

The piece places AIUC in a broader narrative of rising AI risk as models become more capable, underscoring the demand for verifiable safety assurances in enterprise deployments.

AIUC’s testing framework mirrors cybersecurity practices (inspired by SOC 2) and draws input from a consortium of about 250 security and risk leaders who purchase AI agents.

The company closed a $40 million Series A led by Ribbit Capital, with prior seed funding totaling $55 million from investors including Nat Friedman and Anthropic co-founder Ben Mann.

AIUC has developed AIUC-1, an industry-standard framework for evaluating AI agents, applied to around 5,000 risk/attack combinations and used by firms like Cursor, ElevenLabs, Harvey, KPMG, Lovable, UiPath, and Fin to assess jailbreaks, hallucinations, and data leaks.

The safety standard AIUC-1 is paired with a testing service that yields a detailed 100-page safety report after conducting thousands of tests.

Industry context includes calls from leaders like Anthropic’s Dario Amodei to pace frontier AI development and consider embedding third-party evaluators, with METR named as a potential partner.

AIUC intends to audit and recertify models quarterly to reflect evolving threats and plans to extend certification to frontier models powering critical applications.