Fijishi - Not compliance. Not infrastructure. Certifying whether AI decisions hold up before a regulator, a court or an insurer.
The question
Compliance frameworks set out which rules to follow. Technical assurance tests whether a system works. Fijishi asks the question that comes after both: when an AI system makes an underwriting, claims, credit or other consequential decision, can the organisation show which rule the decision met, who was accountable and who could stop it?
How we work
- We measure against a named standard, not opinion.
- Our Index method is published, and every correction is public, dated and logged.
- The Index is independent of who pays: no one pays to appear in it or to change an entry, and we do not sell or pitch to any regulator or government the Index covers.
Certification and review, for firms
We offer certification and review to firms, including insurers and reinsurers, through three divisions. Each engagement is fixed in scope, with the fee fixed at intake.
- Fijishi Constitutional Assurance (FCA): the Kill-Switch Compliance Certificate tests one named AI system against one named override or human-review requirement. The Constitutional Baseline Review assesses one named system, criterion by criterion, against a named standard in force.
- FSDA: the Procurement Pre-Signature Review reads an AI contract for override, IP and upgrade independence.
- Juris: forensic case support when an AI decision is disputed.
Public reference work, free to read
-
Fijishi Jurisdiction Index 2026: AI in Insurance Decisions, published 12 October 2026. 38 entries across 34 jurisdictions; the United States counts as one jurisdiction with five entries. Each entry sets a jurisdiction's own published rules against the published record of AI use in its insurance market, and ends in one sourced question the public record does not yet answer. In 14 entries, the binding rule the entry relies on reaches reinsurers directly. A new edition is due each year on the second Monday of October; the next on 11 October 2027.
PDF edition (CC BY-ND 4.0): 10.5281/zenodo.23222476 - Regulatory Deadline Calendar: dated AI rules for financial-services decisions, insurance first. Each row links its source.
- Research papers and public-consultation submissions since 2023.
Find your jurisdiction
Each entry shows a jurisdiction's own rules, whether the rule it relies on reaches insurers, reinsurers and intermediaries directly, and one open question. A jurisdiction's absence from this edition says nothing about it. How to read an entry: Start here.
Australia ·
Bahrain ·
Bermuda ·
Cameroon ·
Canada ·
Cayman Islands ·
China ·
European Union (EU-wide) ·
Germany ·
Hong Kong ·
India ·
Indonesia ·
Ireland ·
Israel ·
Japan ·
Luxembourg ·
Malaysia ·
Malta ·
Mauritius ·
Nigeria ·
Pakistan ·
Peru ·
Qatar ·
Saudi Arabia ·
Singapore ·
South Africa ·
South Korea ·
Switzerland ·
Taiwan ·
Thailand ·
United Arab Emirates (DIFC) ·
United Arab Emirates (mainland) ·
United Kingdom
United States:
California ·
Colorado ·
Federal ·
New York ·
Texas
Independence
- The fee for every engagement is fixed at intake, before any work begins, and does not depend on the finding. Fijishi does not certify a system it helped design or build.
- No company, regulator or government was told of its Index entry before publication, and none had any input into its tier or band.
- Each Index entry states any commercial relationship Fijishi has with a party named in it.
The Index's independence and right of reply, as published: Who we are, independence and right of reply
Corrections, new evidence and right of reply: indexreport.protocol@fijishi.com · Corrections and new evidence
Start an engagement
Write to protocol@fijishi.com with the AI system, the rule it must meet and any deadline. We reply within 5 business days. Scope and fee are fixed at intake, before any work begins. How an engagement works
Incorporated in May 2023 by Akhil Sharma and Preethi Sharma; based in Pune.