LO
ConstraintCognitive automation2024-11-12
Hong Kong's securities regulator treats AI-generated investment research and advice as high risk, expects a person to review the output before it reaches clients, and holds firms accountable
Financial analystoccupation page →
Event date / reported
2024-11-12
Evidence stage
ConstraintFailure, rollback, regulation or cost is suppressing adoption. Can lower an assessment or widen its uncertainty.
Tasks this bears on
Explaining to the decision-maker
Turning the model into a recommendation the CFO, the partner or the client will act on — and answering their questions live.
Being augmented✓ Evidence-backed
Orchestrating the analytical tooling
Setting up the data feeds, model assistants and checks so the team's output is fast and trustworthy.
New task✓ Evidence-backed
Where this applies
A regulator's circular to licensed firms. It notes firms using AI language models to summarise information, generate research reports and identify investment signals; it considers using them to provide investment recommendations, advice or research to investors or clients a high-risk use case; for such uses it expects model validation and ongoing monitoring, and a human in the loop to review the output for factual accuracy before relaying it to the user; it treats customising, training and calibrating a model as model development needing governance; and it says firms remain accountable for their output regardless of the mitigation adopted. The circular's verb is 'should', presented as requirements; the human checks facts, not the investment judgement.
What this means
In Hong Kong, research or advice drafted by AI does not go straight to a client: a person checks it first, the firm answers for it, and building or tuning the model is itself a governed activity. Explaining to the client stays human-mediated, and setting up the tooling becomes accountable work.
What it does not yet show
One regulator's guidance for licensed firms; it does not show how analysts now communicate or how much of their work AI drafts.
What you can check
Open the SFC circular of 12 November 2024 on generative AI language models and find "having a human in the loop to address hallucination risk".
Does it change the assessment?
No. The impact index is never moved by a single event. Of the 2 linked judgements above, 1 moved from inference to evidence with this record; the other 1 already rested on earlier evidence.
Source
Securities and Futures Commission (Hong Kong) — Circular to licensed corporations, "Use of generative AI language models" (12 November 2024) · verified 2026-09-27 · Claude (VOLO agent) · interpreted 2026-09-27 · Claude (VOLO agent)
Primary source — published by the party that did this, or the authority of record. No co-signature needed.