AI agents are messing up at companies everywhere, and a lot of business insurance was never written for it. Here’s what the gap looks like, who’s already in the market, and three ways to enter this weekend without being an insurance company yourself.
Companies are handing AI their email, customer service, refunds and contracts. When the AI gets it wrong (sends the wrong message, approves a bad refund, leaks private data), the company calls its insurance.
The problem: most policies were never written with AI in mind, and some are now writing it out on purpose. In January 2026, Verisk’s ISO rolled out optional generative AI exclusions for general liability policies. That gap is the opportunity.
01
978% more AI lawsuits
US generative AI lawsuits grew 978% from 2012 to 2025, and 137% in 2025 alone, per Gallagher Re.
02
74% of small businesses use AI
In HSB’s survey of 1,000 small and mid-sized businesses, 74% already use AI and 91% plan to. Most owners have never checked whether their policies cover what that AI does.
03
Cyber insurance is the blueprint
Cyber insurance was a niche add-on in the early 2000s. Munich Re puts it at about $16.3 billion in 2025 and expects it to more than double by 2030. AI risk touches every business that uses AI, which is almost all of them.
Know who’s moving before you enter. There aren’t many yet.
01
Munich Re: aiSure
One of the world’s biggest reinsurers. aiSure covers losses from AI errors, aimed at AI vendors and companies that deploy AI.
02
Armilla
A Lloyd’s coverholder selling affirmative AI liability insurance (model errors, agent mistakes, privacy leaks), with limits up to $25 million per insured.
03
HSB
A Munich Re company. In March 2026 it introduced AI liability coverage for small businesses, added to the business policies of partner carriers (pending regulator approval).
You don’t need to be an insurance company. The whole layer around insurance is open. Pick one.
01
AI risk audits for small businesses
Build a 1-page AI Risk Report. List every AI tool they use, the worst realistic failure for each, and the questions they should take to their broker. A starting price to test: $500 to $2,500 per report. Most owners have no idea where their exposure is.
02
Referral partner for licensed brokers
Team up with one or two licensed brokers who sell business policies and send them businesses with AI exposure. Rules on paying unlicensed referrers vary by state. Many only allow a flat fee that doesn’t depend on a sale, and New York bars fees contingent on placing a policy. Have the broker confirm your state’s rules before you agree on terms.
03
AI governance consulting
Help mid-sized companies document every AI tool they run, who approves what it does, and where a human checks the output. They bring that to their broker and lawyer. This is bigger B2B work (think $5K to $25K projects). Hardest to start, biggest ceiling.
01
Pick your angle
Audits if you like systems. Referrals if you’re a connector. Consulting if you have corporate experience.
02
Pick one niche
Skip “all small businesses.” Pick one industry that uses AI heavily, like e-commerce stores running AI customer service, law firms using AI document review, agencies using AI content tools or clinics using AI scheduling.
03
Build one deliverable
Use the prompt below to draft your 1-page AI Risk Report template for that niche, then make it look good in Canva or Google Docs.
04
DM 5 businesses in your niche
Don’t pitch. Offer the free 1-page report. Use the first one or two conversations to sharpen the offer before you charge.
To sell insurance or get commission on a policy, you need a state license. Some states also license people who give paid advice on coverage, so keep your reports to risk mapping and broker questions, not “buy this policy.” AI insurance is still being defined, so plan to update your template every six months.