Revenue and customer profile
Underwriters price on exposure, and revenue is the simplest proxy for it. They also look at who your customers are. Selling to large regulated enterprises raises the stakes of a failure compared with selling to small teams, and so do customers in healthcare, financial services or other sectors where an error can be expensive.
What your product does
Underwriters ask what decisions your model influences and how much a human reviews its output. A tool that drafts text for a person to edit is a different risk from one that makes an automated decision about credit, hiring or medical triage. Be ready to explain the use cases clearly, the guardrails you maintain and the testing you perform, because clear answers tend to improve terms.
Data and security controls
The data you hold shapes the cyber side of the program. Carriers look for basics such as multi-factor authentication, encryption, endpoint protection, tested backups and a response plan, and for how you handle customer data and training data. Strong controls can meaningfully affect both availability and price, and weak ones can make coverage difficult to place at all.
Limits, retentions and contracts
Higher limits cost more, and a higher retention lowers premium by shifting more of each loss to you. Your customer contracts matter as well. Uncapped liability clauses, broad indemnities and warranties about accuracy increase the exposure that an underwriter sees, and negotiating those terms can help both your legal risk and your insurance cost.
The most reliable way to learn what your own program would cost is to share a short submission with a broker who can approach several markets at once. Quotes will vary between carriers, since appetite for AI risk is not uniform.
Last updated October 22, 2026. Written by the Alton Risk placement team.