Frontier AI company links safety audits to insurance and updated liability rules

A frontier-AI company said it is working on standards, audits and insurance for frontier systems after raising a $40 million Series A. It described an effective audit ecosystem as requiring demand, enough qualified auditors and incentives for auditors to report risks accurately; the near-term bottleneck, it said, is the shortage of world-class frontier-AI auditors.

The company argued that insurers may have stronger incentives than AI labs to identify and report risks because they would bear losses if their assessments were wrong. It said existing liability categories do not neatly cover AI incidents, including cases without clear human intent, and that legal frameworks need updating. Liability also cannot exceed the balance sheets of entities that can be held responsible.

As a possible model for covering risks beyond those balance sheets, the company pointed to mandatory nuclear insurance and mutual coverage among nuclear operators, while cautioning that large AI risks remain difficult to price. It predicted that a combination of labs’ existing exposure, formal insurers and watchdog incentives could be a promising starting point, although major insurers may take time to offer very large-scale coverage.