The Calendar covers AI rules that bear on financial-services decisions: insurance first, across every jurisdiction in the Jurisdiction Index, then banking and credit. Economy-wide AI laws that bind financial firms are included.
It does not cover frontier-model developer laws, rules that bind only government bodies, industry products that are not regulatory instruments, or items with no financial-services application.
What gets a row
A row is added when the instrument meets all four tests below, or is a dated item in a Jurisdiction Index entry:
- names AI, models that include AI, or automated decision-making;
- was issued by a legislature, a regulator or an EU authority;
- is addressed to financial-services firms, or binds firms across the economy; and
- has a printed date of issue, entry into force or application.
Dates
A date is shown only if a source prints it, or if it follows from a period the source prints, running from a printed date; the Status cell then names the provision.
Sources
Every date and status is checked against a primary source - the law, the regulator's own document or the official record - when the row is entered or changed. The source cell gives its type and printed date. The line above each table gives the date of the check; a row checked on another date says so. Where the primary source cannot be read, the cell points to a numbered source note under the table, which says so and names the substitute used.
Status
Being in force is not the same as duties being active. When the two differ, the Status cell says so in words, not with a bare "Active".
Consistency with the Jurisdiction Index
Where an item also appears in a Jurisdiction Index entry, the Calendar uses the same wording, date and status. If the Index and its primary source disagree, the Calendar shows neither until the point is corrected through the Index's corrections policy; the Calendar then follows the corrected Index.
When rows are checked
Each row is checked again whenever it is touched, and at least at each October re-verification of the Jurisdiction Index. The date of the last check is given above each table, or in the row's source note where it differs.
How changes are logged
The Calendar changes only when something real changes - a rule is issued, takes effect, is deferred or withdrawn - never on a schedule. Each change is posted as a dated reply under the status topic it affects. A correction of something we got wrong is also entered in the log below.
Corrections log
| Date | Topic | What changed | Why | Source |
|---|---|---|---|---|
| 10 October 2026 | Currently in force / active | New York: "requires" → "sets out the Department's expectations; calls for" | The circular uses "should" | Regulator circular, July 11, 2024 |
| 10 October 2026 | Currently in force / active . Upcoming | European Union: Article 50 status split by date; "watermarking" → "Article 50(2) marking" | Article 50(2) deferral to 2 December 2026 covers marking only | EU law, OJ 24 July 2026 |
| 10 October 2026 | Currently in force / active . Upcoming | South Korea: fixed date of 22 January 2027 for fines removed; grace period moved to the Status cell | The Ministry stated "at least one year"; no fixed date | Press report, 31 December 2025 |
| 10 October 2026 | Upcoming | Colorado: court stay added; description reworded | Enforcement is paused under a court stay | Attorney General's page |
| 10 October 2026 | Upcoming | Australia: "AI-driven decisions" → "decisions made or informed by computer programs" | Regulator's wording | Regulator release, 30 September 2026 |
| 10 October 2026 | Pending finalization / unresolved | India (RBI): "requires" → "draft"; description reworded | The guidance is a consultation draft | Regulator press release, June 24, 2026 |
Items removed from scope on 10 October 2026: California SB 53 and New York RAISE Act (frontier-model developer laws); California AI Transparency Act, SB 942/AB 853 (generative-AI provider and platform rules); Texas HB 149, TRAIGA (its discrimination provision does not apply to insurance entities providing insurance services, and federally insured financial institutions are deemed compliant; its other provisions have no financial-services application); Canada, Treasury Board Directive on Automated Decision-Making (government-only); Australia, Digital ID private-sector expansion (not an AI rule); Kenya, National AI Policy (no financial-services application); an industry insurance-policy wording standard (not a regulatory instrument); EU AI Act, Annex I high-risk date and new Article 5 prohibitions (no financial-services application); United Kingdom, Artificial Intelligence (Regulation) Bill [HL] (Private Member's Bill at first reading; no stage scheduled).
Corrections and new primary sources: indexreport.protocol@fijishi.com. Last updated: 10 October 2026.