COMPENSATION COMMITTEE POLICY: AI GOVERNANCE INCENTIVE ALIGNMENT (MODELED, NOT GUESSED)
Effective Date: December 25, 2025
Owner: Compensation Committee
Review Cadence: Annual (and upon material AI risk event)
1. Purpose
This policy establishes a board-defensible framework to align executive variable compensation with AI governance, predictability, and evidence-backed decision integrity. The intent is to reward system-building and disciplined governance rather than heroics, narratives, or unverified "AI progress."
2. Scope
Applies to the Annual Incentive Pool (AIP) and, where applicable, Long-Term Incentives (LTI) for covered executives, with primary accountability assigned to the CFO and functional accountability shared across technology, risk, and operations leadership.
3. Policy Statement
Executives will not receive full incentive payouts unless the Company demonstrates:
- (i) Governed AI deployment
- (ii) Predictable modeled outcomes
- (iii) Auditable evidence lineage for material AI-supported decisions
Incentive design must be measurable, reproducible, and verifiable under audit.
4. Definitions (for proxy-ready clarity)
CHI-AI (AI Governance Health Index)
A composite, 0–100 score measuring AI governance system health across evidence integrity, reproducibility, data timeliness, and control adherence.
Modeled Outcome Performance
Performance against pre-declared modeled expectations (e.g., forecast accuracy, variance control, volatility compression) for AI-enabled initiatives.
Governance & System Maturity
Operational control signals that ensure decisions are auditable (e.g., evidence lineage completeness, reproducibility score, data latency).
Material AI System
Any AI model or agent whose failure could materially impact financial reporting, customer outcomes, regulatory exposure, safety, or brand trust.
5. Annual Incentive Pool (AIP) Weighting Framework
| Component | Weight | |-----------|--------| | CHI-AI Score | 40% | | Modeled Outcome Performance | 35% | | Governance & System Maturity | 25% |
Note: Weights may be adjusted by the Committee to reflect business model and risk profile; however, governance integrity (CHI-AI + System Maturity) must not fall below 50% of total variable compensation weighting.
6. KPI-to-Pay Mapping (Board-Defensible Gates)
6.1 CHI-AI Score Tranche (40%) — payout gate
| CHI-AI Score | Payout on CHI Tranche | |--------------|-----------------------| | ≥ 85 | 100% | | 70–84 | Pro-rata (linear) | | < 70 | 0% (narrative risk penalty) |
This gate ensures executives cannot be paid at target while governance integrity degrades.
6.2 Modeled Outcome Performance (35%) — modeled accountability
Measured using pre-declared modeled expectations such as:
- Forecast accuracy and variance control (modeled vs actual)
- Volatility compression (range width reduction)
- Value capture / risk avoided attributable to AI-enabled decisions (where measurable)
6.3 Governance & System Maturity (25%) — control enforcement
- Evidence lineage completeness (source, version, timestamp) for board-facing outputs
- Decision reproducibility score (rerun consistency under identical inputs)
- Data latency (timeliness of inputs supporting model outputs)
7. Governance Controls, Auditability, and Discretion
7.1 Evidence & Auditability
All KPI inputs and calculations must be traceable, versioned, and retained for audit. Material AI decisions must be accompanied by an evidence trail suitable for board review.
7.2 Discretion
Committee discretion may adjust payouts only when documented ex-ante triggers occur (e.g., extraordinary events). Discretion may not override the CHI-AI minimum gate (<70) unless the Committee documents a compelling fiduciary rationale.
7.3 Certification
The CFO and the executive responsible for AI governance (e.g., CIO/CTO/CRO) must certify KPI integrity and control operation at year-end prior to payout approval.
8. Proxy / Public Disclosure Language (Suggested)
Suggested CD&A statement:
The Compensation Committee incorporated AI governance and predictability measures into executive incentive programs to ensure that AI-enabled initiatives are implemented with appropriate controls, auditability, and disciplined performance management. A material portion of annual incentives is contingent upon the Company's AI Governance Health Index, modeled outcome performance (measured against pre-established expectations), and system maturity controls, including evidence lineage completeness, decision reproducibility, and data timeliness.
Appendix A: Example CHI-AI Components
The CHI-AI composite score may include (but is not limited to):
- Evidence Lineage — Completeness of source attribution, version control, and timestamp tracking
- Assumption Explicitness — Clarity and documentation of model assumptions
- Reproducibility — Consistency of model outputs under identical inputs
- Data Latency — Timeliness and freshness of data inputs
- Control Attestation — Documented control testing and sign-offs
Implementation Notes
This policy is designed to be proxy-ready and suitable for inclusion in CD&A (Compensation Discussion & Analysis) sections of proxy statements. It provides:
- Clear, measurable gates that prevent payout when governance fails
- Audit-ready definitions and calculation methods
- Board-defensible rationale for compensation decisions
- Protection against "AI washing" in executive performance narratives
For implementation support or customization for your organization, contact Kincaid RMC.
