Underwriting Superintelligence: Backing Agents you can Sue — Rune Kvist, AIUC
Latent Space ● Covered by 78 sources
AIUC just raised $40M to sell a standard for AI agents and the insurance to back it. The bet: adoption will stall on trust, not model IQ.
Based on reporting by Latent Space — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
AIUC has closed a $40 million Series A led by Ribbit Capital and First Harmonic, and the pitch is plain enough: if AI agents are going to run real work, someone has to be able to answer for the damage when they fail. The company is building AIUC-1, a standard for agent security, safety, and reliability, and it’s pairing that with insurance infrastructure rather than treating trust as a vague branding exercise.
Rune Kvist, AIUC’s cofounder, says the company’s original seed-round thesis has stopped being hypothetical. Back then, the idea was that risk might eventually slow adoption. Now, he argues, risk is the binding constraint. That’s the core of the company’s bet: capability keeps rising, but deployment still runs into liability, responsibility, and the question of who gets sued when an autonomous system goes wrong.
The customer list gives the argument some weight. AIUC says it is working with Cursor, Harvey, Lovable, and ElevenLabs, all of them companies pushing AI into real products instead of demos. And once you’re there, the happy path is not enough. The hard part is stress-testing agents for jailbreaks, hallucinations, and data leaks, because the failure mode isn’t theoretical anymore.
Kvist’s path into all this runs through the early Anthropic days, when the company had around 40 people and was already wrestling with whether to deploy models, make revenue, and how to engage with the outside world. He says Anthropic thought far more seriously than most about what scaling would mean years ahead, including the compute required and the social fallout. That early conviction is now being turned into a business around standards, audits, and insurance.
The larger argument is that AI adoption will not be won by raw capability alone. Governments, frontier labs, and enterprise buyers are all staring at the same trust gap, and AIUC wants to sit in the middle of it. If the company is right, AI infrastructure won’t just be chips and models. It’ll also be the people and policies standing behind the agents when something breaks.
My take — AI-written commentary, not fact-checked reporting
This is the part of AI everyone keeps trying to postpone: accountability. If agents are going to touch money, code, and safety-critical work, then “trust us” is not a control system, it’s a punchline. The funniest possible future is still the least likely one: the industry discovering it needs adults after shipping like teenagers.
Read more about this at: Latent Space