# Faultprint Assessment Draft: Allstate

## Status

Draft generated from the v14 consistency-clean baseline. v14 applies source-stress handling, null-row controls, low-resemblance caveats, and boundary-hygiene fixes.

This draft is findings-only. It does not recommend actions, predict failure, assert illegality, or claim that the institution has failed.

## Corpus Position

- Institution: Allstate
- Sector: Insurance
- Client-eligible findings included: 4
- Caveated findings included: 3
- Findings with source-verification limitations: 1

## Pathology Mix

- Multi-Actor Handoff Diffusion: 2
- AI Authority Acceleration: 2

## Resemblance Mix

- low_resemblance: 1
- moderate_resemblance: 3

## Findings

### Finding 1: AI Authority Acceleration

Condition: AI feature authority publicly limited

Resemblance band: moderate_resemblance

Output status: client_ready

Finding statement: Operating row shows a condition also present in postmortem libraries, but current public evidence is not enough to treat it as high exposure.

Evidence anchor: AI Features are provided for informational and assistance purposes only and do not have authority to bind Allstate.

Source: https://www.allstate.com/terms

Falsification evidence: Internal policies showing website AI outputs are used as binding determinations would weaken this mitigant.

Boundary: Applies to website AI Features; does not establish internal AI operations.

### Finding 2: AI Authority Acceleration

Condition: Personal information may be used with AI

Resemblance band: moderate_resemblance

Output status: client_ready_with_caveat

Finding statement: Operating row shows a condition also present in postmortem libraries, but current public evidence is not enough to treat it as high exposure.

Evidence anchor: Allstate Online Privacy Statement states it may use personal information with AI technology to improve or provide services, develop products, features, and technologies, and prevent fraud or misuse.

Source: https://www.allstate.com/privacy-center/aic-privacy-statement

Falsification evidence: Records showing AI use excludes claims/pricing/underwriting/fraud decision support would downgrade exposure.

Boundary: Does not establish how AI is used in any specific underwriting, claim, fraud, or service decision.

Source verification note: Official source token-limited in automation; human source confirmation required before client use.

### Finding 3: Multi-Actor Handoff Diffusion

Condition: Claims-related third-party sharing

Resemblance band: moderate_resemblance

Output status: client_ready_with_caveat

Finding statement: Operating row shows a condition also present in postmortem libraries, but current public evidence is not enough to treat it as high exposure.

Evidence anchor: Allstate privacy materials state personal information may be shared with companies involved in insurance transactions, including independent claims adjusters, repair shops, and other claims-related companies.

Source: https://www.allstate.com/resources/Allstate/attachments/about/allstate-privacy-statement-2022.pdf

Falsification evidence: Vendor register showing claims-related recipients do not provide AI or automated decision support would downgrade exposure.

Boundary: Does not establish improper disclosure, data-quality failure, or claim harm in any particular claim.

### Finding 4: Multi-Actor Handoff Diffusion

Condition: Supplier AI disclosure required

Resemblance band: low_resemblance

Output status: client_ready_with_caveat

Finding statement: Operating row is visible as an operating condition but has limited postmortem resemblance in this pass.

Evidence anchor: Suppliers must disclose if technology or services involve or incorporate AI systems or models that process Allstate confidential information.

Source: https://www.allstate.com/resources/allstate/attachments/procurement/supplier_code_ethics.pdf

Falsification evidence: Audit records showing supplier AI disclosure is incomplete or not enforced would weaken this mitigant.

Boundary: Does not prove disclosure quality or enforcement.

## Internal Exclusions

No internal revise/hold rows were included in this report draft.

## Report Boundary

This draft is based on the reviewed corpus and current Faultprint evidence libraries. It identifies structural conditions and evidence requirements. It is not a legal opinion, audit conclusion, compliance certification, recommendation, or prediction.