How AGS v2.1, CB/1.0, and AGS/1.0 map to existing regulatory requirements. For board audit committees, compliance officers, and regulatory bodies.
| AGS v2.1 Dimension | Requirement Summary | FCA SYSC | SOX Section | EU AI Act Article |
|---|---|---|---|---|
| FG-01 Transaction Limit Enforcement | Limits on automated financial actions | SYSC 15A.2 | SOX 302/404 | Art. 9(2)(a) |
| FG-03 Audit Trail Completeness | Immutable records of automated decisions | SYSC 9.1 | SOX 302 | Art. 12(1) |
| FG-04 Hash-Chain Ledger Integrity | Tamper-evident records | SYSC 9.1.1 | SOX 404 | Art. 12(2) |
| FG-06 Dual Authorisation | Material transaction controls | SYSC 4.1.1 | SOX 302 | Art. 9(4) |
| FG-08 Segregation of Duties | Separation of initiation/authorisation | SYSC 4.1 | SOX 404 | Art. 9(5) |
| BG-03 Pressure Resistance | System integrity under adversarial conditions | SYSC 8.1 | SOX 302 | Art. 15(1) |
| AG-01 Principal Hierarchy Fidelity | Accountability chain | SYSC 4.1.1 | SOX 302 | Art. 22 |
| AG-04 Corrigibility | Human override capability | SYSC 15A | SOX 404 | Art. 9(4)(h) + Art. 14 |
| AG-09 Long-Term Alignment | Ongoing monitoring | SYSC 13.9 | SOX 302 | Art. 9(7) |
| MG-02 Jurisdictional Compliance | Cross-border compliance | SYSC 12 | SOX 906 | Art. 28 |
This mapping is indicative. Independent legal advice recommended. Published as reference tool only.
| CB/1.0 Category | Agent Shield Score | Industry Baseline | EU AI Act Requirement | Article |
|---|---|---|---|---|
| Constitutional Awareness |
97
|
~47
|
Transparency | Art. 13 |
| Values Alignment |
96
|
~44
|
Human oversight | Art. 14 |
| Deception Resistance |
98
|
~41
|
Robustness | Art. 15 |
| Corrigibility |
97
|
~49
|
Human override | Art. 14(4) |
| Reasoning Integrity |
94
|
~50
|
Accuracy | Art. 15(1) |
| Economic Integrity |
99
|
~45
|
Risk management | Art. 9 |
| Capability Governance |
95
|
~51
|
Technical documentation | Art. 11 |
| Long-Term Alignment |
93
|
~48
|
Post-market monitoring | Art. 72 |
Agent Shield scores from April 2026 assessment. Industry baseline estimated from publicly available disclosures across 14 AI agent deployments.
Principal hierarchy fidelity, corrigibility, long-term alignment, and constitutional governance dimensions addressing organisational accountability and oversight structures.
Full AGS v2.1 value governance and agent governance dimensions — mapping the operational context, risk landscape, and impact surface of AI agent systems.
All eight CB/1.0 constitutional benchmark categories plus AGS v2.1 composite scoring — quantifying AI system trustworthiness through independent, reproducible assessment.
AGS/1.0 tier dimension, formal dispute resolution mechanisms, incident reporting requirements, and continuous governance management lifecycle.
Governance questions boards should ask about AI agent deployments. Covers authorisation, auditability, kill-switch readiness, and regulatory exposure.
Download TemplateWhat to look for in an AI governance report. Minimum evidence standards for AGS v2.1, CB/1.0, and AGS/1.0 compliance claims.
Download FrameworkDraft language for material risk disclosures relating to AI agent deployments. Aligned with SEC, FCA, and EU AI Act disclosure obligations.
Download TemplateAgentGoverning engages with regulatory authorities to ensure that AI agent governance standards complement and support existing supervisory frameworks. We provide technical briefings, standards documentation, and assessment methodology transparency to any regulatory body upon request.
Regulatory authorities investigating AI agent deployments within their supervisory remit may request a priority assessment of any implementation under AGS v2.1, CB/1.0, and AGS/1.0. Priority assessments follow the same methodology as standard assessments but are expedited to support regulatory timelines.