Brazil / automated-systems liability 2026Q3 / r5

First-case jurisdiction

See the loss signal before it becomes the loss history.

IISA turns Brazil’s public court, complaint, and enforcement record into a quarterly decision instrument for teams pricing and governing automated-system risk.

Built for Insurers & reinsurers Global deployers Risk, legal & compliance

The buying case

One instrument. Three consequential decisions.

The same measured event stream changes the verb depending on who is buying.

01

Risk capital

Price

Build Brazil-specific frequency, severity, and wording assumptions before credible loss histories arrive.

02

Enterprise exposure

Govern

Translate observed failures into evidence-readiness, escalation, and deployment requirements.

03

Legal & compliance

Prepare

Benchmark outcomes, technical-evidence patterns, and collective or regulatory escalation.

What the instrument measures

From public record to decision-ready loss intelligence.

Every quarter, IISA classifies the Brazilian record into boring, decomposable indicators that actuaries, risk owners, and regulators can interrogate. No composite score. No hidden weighting.

Public-source spine
Courts / complaints / regulators
Release rhythm
Quarterly / dates published ahead
01 FREQ

Incident frequency

Sector × state × court level

raw and denominator-aware
02 SEV

Damages distribution

Median, p90, moral / material

reported coverage retained
03 OUT

Outcome rates

Favorable, partial, dismissed

court posture by cohort
04 LOG

Evidence production

Demanded, produced, absent

the log-gap indicator
05 AGT

Agentic share

Segment level + quarter-on-quarter

leading curve, not hype
06 ENF

Escalation

Collective and administrative actions

correlated-defect signal

A working release, not a pitch-deck mockup

Inspect the instrument’s public surface.

This signed synthetic quarter demonstrates the delivery format, denominator discipline, assessor QA, and reproducibility contract.

Fabricated proof data
Decisions
48
fabricated n = 48
Sectors
4
represented
Favorable / partial
41.7%
20/48
Median reported damages
R$3,100
45/48 reported
Synthetic decisions grouped by sector
The chart is a delivery specimen. Values are fabricated.

Conflict-clean commercial lanes

Different buyers. Deliberately different engagements.

The lane architecture protects the neutrality of the measurement layer while letting the same asset answer distinct business questions.

01 SVC

Services to risk capital

Insurers / reinsurers / MGAs / investors

Need
Price a market with scarce local loss history.
Offer
Exposure modelling, underwriting-criteria design, and bespoke portfolio or diligence cuts.

Private book data is never published.

02 PRD

Products to operators

Global deployers / banks / fintech / retail

Need
Understand and govern Brazilian consumer exposure.
Offer
Quarterly index access, annual exposure briefings, and evidence-readiness requirements.

No defense retainers as core revenue.

03 PUB

Public data to institutions

Regulators / researchers / public-interest bodies

Need
Allocate attention using an inspectable fact base.
Offer
Public indicators, methods, and aggregate briefings.

No pay-to-rank and no privileged findings.

Engagement formats

Buy the decision surface you need.

Each format is a different cut of the same measured event graph, so bespoke work compounds the instrument instead of creating a services swamp.

A

Annual / quarterly delivery

Index subscription

Core indicators, release notes, signed data pack, and buyer-specific brief.

Core offer
B

One-off / standardized

Exposure or diligence cut

Sector, target, or transaction exposure translated into a decision memo.

Demand-gated
C

Live / maximum once per quarter

Underwriter briefing

A working session on priors, wording gaps, and what to ask the insured.

Demand-gated
D

Licensed / feed-ready

Underwriting questionnaire

A reusable evidence-readiness form linked to signed indicator definitions.

L1 extension
Discuss fit and scope

Why this can become infrastructure

The defensibility is in the measurement contract.

Anyone can publish a view on AI liability. The durable asset is a regenerable event history, a stable taxonomy, measured assessor performance, and a release cadence the buyer can rely on.

I

Brazil-native source reading

Portuguese court text and local consumer-liability doctrine classified into insurer-mappable fields.

II

Measured judgment

Blind and assisted assessments remain distinct; coverage and agreement publish field by field.

III

Signed publication

Every public artifact and replay input is committed in a detached-signature manifest.

IV

Decomposable outputs

No scalar risk score. Counts, rates, uncertainty, and missingness remain visible.

The compounding asset

Measure once. Serve at increasing leverage.

NOW Measured events

Quarterly loss instrument

NEXT Signed feeds

Indicator and event APIs

GATED Work-history profiles

Methodology-published reputation

GATED Market interfaces

External insurance, bounty, and escrow rails

Dover Intel records, verifies, and measures. It does not custody funds or execute judgments.

Technical receipt

Verify this publication.

The commercial page and the machine-readable artifacts are part of the same signed release.

Release ID
iisa.synthetic:2026Q3:r5
Generated
Composite score
Not emitted
Production gate
Not passed
Show release identifiers and assessor scorecards
Methodology SHA-256
95e332abb10f5d62c5131444a6e94084c862b5af73d9a126ae602f7ca4a792bf
Input snapshot SHA-256
5ff3e9018f546fe097d8d8d7edecb86e1ec3c9bcf9852f4c68257f74c3299056

Assessor QA

$.payload.system_type 16 eligible 15 answered 90.8% exact agreement
$.payload.tech_evidence 16 eligible 16 answered 91.7% exact agreement

Founding cohort

If Brazil touches your AI exposure, start with the briefing.

Bring the portfolio, deployment, or policy question. We will show which parts the instrument can answer now, which require a bespoke cut, and which should remain unknown.

Request the briefing