# Faultprint Structural Assessment: Workday Responsible AI

## Assessment Boundary

This assessment identifies document-visible structural conditions. It does not assert institutional failure, illegality, misconduct, causation, or future outcomes.

## Corpus Position

|field|value|
|---|---|
|Institution|Workday Responsible AI|
|Sector|HR technology|
|Corpus reviewed|Public source materials represented in the current operating library.|
|Assessment scope|Selected source-anchored operating rows for example assessment formatting.|

## Module Screen

|module|screen result|reason|
|---|---|---|
|Contestability Lag|visible condition|Alternative procedures and human review are described|
|AI Authority Acceleration|visible condition|Fairness efficacy robustness and monitoring controls|
|Record-Authority Capture|not visible in selected rows|No source-anchored finding selected for this module.|
|Access Interruption Before Review|not visible in selected rows|No source-anchored finding selected for this module.|
|Documentation Burden Transfer|not visible in selected rows|No source-anchored finding selected for this module.|
|Classification Gatekeeping|not visible in selected rows|No source-anchored finding selected for this module.|
|Evidence-Container Illusion|not screened as standard module|Pilot module; use only with strong source support.|
|Multi-Actor Handoff Diffusion|not screened as standard module|Pilot module; use only with strong source support.|
|Deadline Compression|not screened as standard module|Pilot module; use only with strong source support.|

## Findings

### FP-F-001: Risk assessments before use

|field|content|
|---|---|
|Module|Contestability Lag|
|Condition name|Risk assessments before use|
|Source anchor|Workday states AI features receive risk assessment reviews during development and before use.|
|Source location|https://www.workday.com/en-us/artificial-intelligence/responsible-ai-practices.html|
|Why it matters structurally|Operating row shows a condition also present in postmortem libraries, but current public evidence is not enough to treat it as high exposure.|
|Boundary|Does not prove risk reviews prevent harm.|
|Falsification evidence|Missing or superficial assessment artifacts would weaken this mitigant.|

### FP-F-002: Alternative procedures and human review are described

|field|content|
|---|---|
|Module|Contestability Lag|
|Condition name|Alternative procedures and human review are described|
|Source anchor|Workday describes alternative procedures including human review and optionality for data subjects.|
|Source location|https://www.workday.com/en-us/artificial-intelligence/responsible-ai-practices.html|
|Why it matters structurally|Operating row shows a condition also present in postmortem libraries, but current public evidence is not enough to treat it as high exposure.|
|Boundary|Does not prove alternatives work in practice.|
|Falsification evidence|Records showing alternative review is hard to access or ineffective would weaken this mitigant.|

### FP-F-003: Worker predictions and categorizations create consequential exposure

|field|content|
|---|---|
|Module|AI Authority Acceleration|
|Condition name|Worker predictions and categorizations create consequential exposure|
|Source anchor|Workday states risk tiers consider predictions or categorizations about individual workers and economic opportunities.|
|Source location|https://www.workday.com/en-us/artificial-intelligence/responsible-ai-practices.html|
|Why it matters structurally|Operating row shows a condition also present in postmortem libraries, but current public evidence is not enough to treat it as high exposure.|
|Boundary|Does not prove discriminatory impact or adverse decisioning.|
|Falsification evidence|Customer records showing no consequential worker-facing use would downgrade exposure.|

### FP-F-004: Human final decision maker for critical decisions

|field|content|
|---|---|
|Module|AI Authority Acceleration|
|Condition name|Human final decision maker for critical decisions|
|Source anchor|Workday states humans remain final decision makers for critical decisions.|
|Source location|https://www.workday.com/en-us/artificial-intelligence/responsible-ai-practices.html|
|Why it matters structurally|Operating row is visible as an operating condition but has limited postmortem resemblance in this pass.|
|Boundary|Does not prove meaningful human review.|
|Falsification evidence|Decision logs showing rubber-stamp or automated finality would weaken this mitigant.|

### FP-F-005: Fairness efficacy robustness and monitoring controls

|field|content|
|---|---|
|Module|AI Authority Acceleration|
|Condition name|Fairness efficacy robustness and monitoring controls|
|Source anchor|Workday describes fairness efficacy robustness testing scheduled testing maintenance traceability and monitoring.|
|Source location|https://www.workday.com/en-us/artificial-intelligence/responsible-ai-practices.html|
|Why it matters structurally|Operating row is visible as an operating condition but has limited postmortem resemblance in this pass.|
|Boundary|Does not disclose actual test results.|
|Falsification evidence|Failed or non-remediated testing records would weaken this mitigant.|

## Evidence Ledger

|evidence id|document/source|anchor|supports finding|
|---|---|---|---|
|E-001|https://www.workday.com/en-us/artificial-intelligence/responsible-ai-practices.html|Workday states AI features receive risk assessment reviews during development and before use.|FP-F-001|
|E-002|https://www.workday.com/en-us/artificial-intelligence/responsible-ai-practices.html|Workday describes alternative procedures including human review and optionality for data subjects.|FP-F-002|
|E-003|https://www.workday.com/en-us/artificial-intelligence/responsible-ai-practices.html|Workday states risk tiers consider predictions or categorizations about individual workers and economic opportunities.|FP-F-003|
|E-004|https://www.workday.com/en-us/artificial-intelligence/responsible-ai-practices.html|Workday states humans remain final decision makers for critical decisions.|FP-F-004|
|E-005|https://www.workday.com/en-us/artificial-intelligence/responsible-ai-practices.html|Workday describes fairness efficacy robustness testing scheduled testing maintenance traceability and monitoring.|FP-F-005|

## Counter-Signals

|counter-signal|source location|effect on finding|
|---|---|---|
|Workday describes alternative procedures including human review and optionality for data subjects.|https://www.workday.com/en-us/artificial-intelligence/responsible-ai-practices.html|May narrow or weaken the condition if confirmed by operating records.|
|Workday states humans remain final decision makers for critical decisions.|https://www.workday.com/en-us/artificial-intelligence/responsible-ai-practices.html|May narrow or weaken the condition if confirmed by operating records.|

## Falsification Checklist

|finding|records that would weaken or rebut the finding|
|---|---|
|FP-F-001|Missing or superficial assessment artifacts would weaken this mitigant.|
|FP-F-002|Records showing alternative review is hard to access or ineffective would weaken this mitigant.|
|FP-F-003|Customer records showing no consequential worker-facing use would downgrade exposure.|
|FP-F-004|Decision logs showing rubber-stamp or automated finality would weaken this mitigant.|
|FP-F-005|Failed or non-remediated testing records would weaken this mitigant.|

## Non-Claims

- This assessment does not recommend institutional action.
- This assessment does not predict failure.
- This assessment does not assert illegality.
- This assessment does not score the institution against a public benchmark.