Anthropic Alum's AIUC Raises $40M to Insure Against Rogue AI Agents

A startup founded by an early Anthropic employee and the former chief operating officer of METR has raised $40 million to build an insurance business around the risks of autonomous AI agents. The company, Artificial Intelligence Underwriting Company, closed the Series A round led by Ribbit Capital, with First Harmonic also participating. The bet is straightforward: as AI agents take actions in the real world, someone will need to price and absorb the losses when they go wrong.

AIUC's founding team pairs hands-on model-development experience with safety-evaluation operations. One founder worked at Anthropic in its early days, while the other ran operations at METR, the nonprofit known for testing whether frontier models can deceive, self-replicate or evade oversight. That combination matters for an insurer, because underwriting AI risk requires both technical fluency and a method for measuring failure modes that traditional actuarial tables do not capture.

The company's premise is that AI agents differ from ordinary software in ways that make standard corporate policies inadequate. An agent that books travel, moves money or negotiates contracts can cause harm through a chain of individually reasonable decisions, leaving liability unclear. AIUC aims to write coverage that maps those exposures and, in the process, impose underwriting standards on how agents are built and deployed.

The $40 million round is modest by frontier-model standards, where single training runs can cost more than the entire raise. Its significance lies in the category rather than the check size. Insurance is one of the few mechanisms that can turn abstract safety concerns into concrete financial incentives, because a carrier that refuses coverage or charges a steep premium changes a developer's economics overnight.

Ribbit Capital's involvement signals that investors see agent reliability as an infrastructure market rather than a research curiosity. First Harmonic's participation adds a second institutional backer to the cap table. Neither firm has disclosed the valuation or the specific terms of the round.

What to watch is whether AIUC can actually quantify agent risk before a major incident forces the issue. The company has not said which agent developers it will cover, what premiums it will charge, or how it will handle claims when responsibility is split across a model provider, an orchestration layer and an end user. Those details will determine whether this becomes a real insurance market or a well-funded pilot.