A day after Anthropic researcher Jacob Coxon left his job due to concerns that artificial intelligence (AI) could pose a global threat by the end of the decade, the founders of a new AI safety startup, Rune Kvist and Rajiv Dattani, discussed their efforts to help businesses safely use AI. Kvist, who worked early on at Anthropic and is married to Dattani’s sister, said that as AI systems become more advanced, they are also harder to manage and control. Dattani, who previously worked as COO at METR, an AI safety research group, co-founded the Artificial Intelligence Underwriting Company (AIUC). The startup aims to help companies ensure their AI systems are safe and reliable. AIUC has already secured clients like Cursor, Lovable, Harvey, and ElevenLabs. Recently, the company announced it had raised $40 million in a Series A funding round, led by Ribbit Capital, with additional support from First Harmonic. Earlier, the startup had raised $15 million in a seed round from various investors, including Anthropic co-founder Ben Mann, bringing its total funding to $55 million. AIUC has developed a method to test and certify AI systems, similar to how cybersecurity standards like SOC 2 are used to evaluate the safety of digital systems. The company created its own standard called AIUC-1 and a testing process to assess AI agents against it. To shape these tests, AIUC gathered around 250 leaders in security and risk management—people who purchase AI tools—to identify what they look for in an AI system and what issues they want addressed. Using this input, AIUC runs an AI agent through thousands of tests to see how it behaves in situations like being tricked into doing something harmful (jailbreaks), making up false information (hallucinations), or leaking sensitive data. The results are compiled into a detailed report that highlights where the AI performs safely and where it falls short. While AIUC uses AI to run the tests and analyze the data, human experts review and confirm the final audit. Dattani’s former employer, METR, which he led as COO in 2024 and 2025, also conducts similar safety evaluations, but its focus has been more on performance than on security. METR was one of the groups that OpenAI consulted after an incident involving the AI platform Hugging Face. Anthropic CEO Dario Amodei has called for the AI industry to slow down the development of advanced systems, noting a rise in incidents where AI behaves unpredictably or dangerously. Amodei suggested that companies developing cutting-edge AI should use independent third-party evaluators to check their systems’ safety, naming METR as one possible partner. While AIUC does not plan to embed its evaluators directly into customer organizations, its overall goal is similar: to give businesses an independent assessment of how safe their AI systems are.