Data prices · Insurance
What AI labs pay for insurance data
Claims files, underwriting notes and adjuster decisions are training data AI labs can't find in public sources. No public prices exist for licensing claims data to AI labs, and state insurance privacy rules shape what an insurer may share.
What buyers want
Records your team already keeps
01
Claims journeys
First notice to settlement, with each decision and the reason given.
02
Underwriting decisions
What was quoted, declined or referred, and why.
03
Fraud and recovery outcomes
Which flags were right and how recoveries turned out.
Price signals
What public deals and research show
$200-$2,000
Typical price per checkable task. A claim with a known outcome can become one.
Epoch AI, 2026We found no public price for licensing claims files for AI training. Industry claims databases are sold as services on negotiated contracts, not at published rates.
The wider market
$203M
Data licensing contracts Reddit disclosed in early 2024, with terms of two to three years.
TechCrunch$250M+
Reported value of News Corp's five-year content deal with OpenAI, including cash and credits.
Press Gazette$10k-$100k
Per-company payouts for shut-down startups' Slack messages, tickets and email archives, across nearly 100 deals.
Forbes, April 2026$200-$2,000
Typical price per reinforcement learning task, according to Epoch AI's interviews with people who build them.
Epoch AI, 2026Pricing factors
What moves the price
- Decision trails
- Claims systems keep event histories, which become episodes of adjuster decisions.
- Outcomes
- Settlement amounts (as ranges), litigation and recoveries make the data checkable.
- Line of business
- Specialty lines with less shared industry data are worth more.
- Years of history
- Longer runs of records show change over time and how outcomes played out, which a single year can't.
- Rights you can prove
- Buyers pay for data you own or are allowed to license. Data held for customers without their permission won't sell, so our calculator cuts the estimate when rights are unclear.
Our calculator
How we estimate insurance data
Our indicative model starts from a base value per system for about five years of history at an 11-50 person company, then applies a sector weight of 1.3x, your history, team size and rights. It is an estimate we confirm on a call, not a market price.
| System | Base value |
|---|---|
| Claims & underwriting files | $50k |
| Docs, wikis & SOPs | $10k |
| Call recordings & transcripts | $35k |
Recurring
Paid again every month
New claims open and close every month. Each month's closed claims can go out as a batch on a refresh term.
How recurring revenue worksRules that apply
Insurers share consumer data under federal financial privacy rules and state insurance privacy laws based on NAIC models, and regulators expect governance over AI and third-party data.
Source: NAIC, big data and AIValuation
What is your data worth?
No access to your systems needed. We confirm the number on a 30-minute call.
Questions
Insurance data, answered
Can an insurer share claims data at all?
What about health information in claims?
Will this help competitors?
Keep reading
Selling insurance dataSelling insurance claims data to AI companies
Claims files are rare training data for AI labs, but insurance privacy rules come first. What buyers want, what must be removed, and how licenses work.
5 min read
CCPA and selling data
Under the CCPA, licensing personal information for money is a sale. Data that meets California's three-part deidentified test is not personal information at all.
5 min read
k-anonymity in plain English
A dataset is k-anonymous when every record looks identical to at least k-1 others on the fields an outsider could know. Here is how it works and where it fails.
5 min read
Prices in other industries