CriticalMisadvice◐Demonstrated
Unlicensed / unsuitable advice
Content Safety & IntegrityDescription
The model gives investment, tax, or credit advice that is unsuitable, unlicensed, or breaches conduct rules.
Example scenario
A bot gives specific stock recommendations without an advisory licence, breaching SEBI conduct rules.
Real-world evidence◐Demonstrated
SEC examination sweeps (2021–2023) of robo-advisors found suitability assessment deficiencies in algorithmic systems, and academic work has demonstrated that recommendation engines can systematically ignore financial circumstances. However, a confirmed enforcement action specifically attributing suitability failure to an AI model—rather than to broader firm process failures—has not been publicly documented at scale.
Tested by FinProof
Investment advice
Benchmarked by FinProof Primary mitigations
- Advice guardrails & scope limits
- suitability checks
- mandatory disclaimers
- human-in-the-loop for advice
- conduct review.
Detection signals
Advice-classification monitoring; suitability-rule checks; sampled conduct review.
Mitigating controls
7 Non-agentic controls
ZYC-SEC-001Non-Agentic
Adversarial Testing
Security
ZYC-ETH-001Non-Agentic
Non-Maleficence
Ethical Governance
ZYC-ACCU-001Non-Agentic
Lexical Similarity
Accuracy
ZYC-LEGAL-001Non-Agentic
AI Regulation Compliance
Legal Compliance
ZYC-LEGAL-005Non-Agentic
Advice Boundary
Legal Compliance
ZYC-ETH-002Non-Agentic
Autonomy Preservation
Ethical Governance
ZYC-ETH-003Non-Agentic
Beneficence
Ethical Governance