Hong Kong AI Risk Tiers for Marketing
Tags: Frameworks, Regulatory
TL;DR —
🔗 Synthesized insight
- Classify chatbots, creative tools, targeting, and hiring aids under HK’s four‑tier system to right‑size controls and sign‑offs.
- Use AAIA reviews and human oversight gates where delivery outcomes or identity content raise stakes.
Why it matters for HK marketers: Right tiering prevents over‑ or under‑controlling AI—saving time while keeping high‑impact uses safe and approvable.
Mapping common marketing uses
Unacceptable risk (prohibited)
- Subliminal manipulation or safety‑threatening uses; deepfake misuse to impersonate real individuals for deception.
High risk (conformity + HITL + real‑time monitoring)
- Rare in standard marketing; treat identity/financial artifacts (e.g., synthetic ID documents, financial materials) as high‑risk content if created—apply watermarking/traceability and stringent oversight.
Limited risk (transparency, opt‑outs, audits)
- Customer chatbots that discuss prices, eligibility, or sensitive scenarios—deploy clear disclosures, human‑in‑the‑loop escalation, and output monitoring.
- Recruitment/assessment aids used by HR/TA functions tied to marketing orgs—ensure transparency, opt‑outs, and auditability.
Low risk (self‑certification)
- Creative assistants for copy/imagery; internal brainstorming; ad variants—maintain content safety checks and IP provenance.
Governance moves that unlock approvals
- AAIA + Risk Gating: Trigger procurement question sets for third‑party tools; escalate limited/high‑risk to PSC/PAT and IT Board/CIO as required.
- Human oversight placement: Add review gates at deployment/usage for chatbots and any identity/finance collateral.
- Traceability for high‑risk outputs: Watermark or embed codes; verify sources; block publishing when verification fails.
- Liability awareness: Treat chatbot copy as binding; misstatements can create liability.
4-tier risk framework: Unacceptable, High, Limited, Low.
3 high‑risk output examples cited: deepfakes, ID documents, financial materials.
So what for marketers
Tier every AI use at intake, then align evidence: disclosures and HITL for limited risk; watermarking and senior approvals for any identity/financial content; streamlined self‑cert for low‑risk creative tools.
---
This page was synthesized by AI from themes across multiple member contributions, rather than extracted from a single source document. It may contain interpretive connections or inaccuracies; verify key claims against the source pages before citing.
Sources:- Synthesis — cross-content analysis
← Back to Knowledge Base