evidoria

← Back to browse

Good practice Imported

GenA.I. Sandbox++ — Hong Kong's Four-Regulator Agentic-AI Testing Programme for Banks, Insurers and Asset Managers

Hong Kong · Hong Kong · See the Hong Kong profile · See the Hong Kong profile

Evidence: Descriptive / self-reported Top 12% 80/100 · Ask Evidence Copilot about this practice

Hong Kong's four financial regulators (HKMA, SFC, IA, MPFA) expanded 2024's bank-only GenA.I. Sandbox into GenA.I. Sandbox++, whose first cohort (Aug 2026) selected 36 agentic-AI use cases from nearly 100 proposals across 30 institutions and 27 tech partners.

15 use cases
Use cases selected, 2024 cohort (Dec 2024)
36 use cases
Use cases selected, 2026 Sandbox++ cohort (Aug 2026)
30 institutions
Financial institutions participating, 2026 cohort (Aug 2026)
27 partners
Technology partners, 2026 cohort (Aug 2026)
GenA.I. Sandbox++ — Hong Kong's Four-Regulator Agentic-AI Testing Programme for Banks, Insurers and Asset Managers

Details

Maturity
Scaling
Promoter
Hong Kong Monetary Authority (HKMA), Securities and Futures Commission (SFC), Insurance Authority (IA) and Mandatory Provident Fund Schemes Authority (MPFA), with Hong Kong Cyberport
Period
August 2024–ongoing (Sandbox++ first cohort announced August 2026; technical trials from late 2026)
Keywords
financial regulation, agentic AI, risk management, anti-fraud, banking, insurance, asset management

Context

The Hong Kong Monetary Authority (HKMA) and Cyberport launched the original GenA.I. Sandbox on 13 August 2024 (formal circular 20 September 2024), with an inaugural cohort (19 December 2024) selecting 15 generative-AI use cases from over 40 proposals involving 10 banks and 4 technology partners.

Objectives

On 5 March 2026, HKMA was joined by the Securities and Futures Commission (SFC), the Insurance Authority (IA) and the Mandatory Provident Fund Schemes Authority (MPFA), together with Cyberport, to launch GenA.I. Sandbox++, widening the programme from banking alone to banking, securities and capital markets, asset and wealth management, insurance, mandatory provident funds and stored-value facilities.

Activities

The first Sandbox++ cohort, announced 27 August 2026, selected 36 agentic-AI use cases from nearly 100 proposals, involving 30 financial institutions and 27 technology partners, organised around risk management, anti-fraud measures and customer experience. Participants are onboarded to a platform run on Cyberport's AI Supercomputing Centre, with technical trials due to start later in 2026.

Results

The announcement documents selection scale (36 use cases, 30 institutions, 27 tech partners) but as of these sources, technical trials had not yet started, so no performance, safety or consumer-outcome results are available for this cohort. The clearest evidence so far is the programme's growth from 15 use cases/10 banks in 2024 to 36 use cases/30 institutions across four regulators in 2026.

Implementation
Indicative cost
High (€500k–€5M) — Not publicly disclosed; involves shared regulator/Cyberport infrastructure investment (AI Supercomputing Centre) across four regulators and dozens of institutions.
Time to results
Medium (1–3 years) — Original Sandbox launched August 2024; Sandbox++ launched March 2026; first Sandbox++ cohort announced August 2026 with technical trials starting later in 2026.
Staffing & skills
Hong Kong Monetary Authority (HKMA), Securities and Futures Commission (SFC), Insurance Authority (IA) and Mandatory Provident Fund Schemes Authority (MPFA) as joint regulators, Hong Kong Cyberport, providing shared infrastructure (AI Supercomputing Centre), 30 participating financial institutions and 27 technology partners

Conditions for success

  • A competitive selection process narrowing nearly 100 proposals to 36 use cases
  • Joint governance across four regulators with named accountable leadership at each authority
  • Shared technical infrastructure (Cyberport's AI Supercomputing Centre) so institutions do not each build their own testing environment

Common failure modes

  • As of the sources reviewed, technical trials had not yet started, so no performance, safety or consumer-outcome evidence exists for this cohort

Where it fits

Governance type
national/regional financial regulators (Hong Kong SAR)
Scale
sector-wide (banking, securities, insurance, pensions)
Income level
high-income

Commonly funded by

National / regional programmes

Indicative funding routes for practices of this type — always check each programme's current calls and eligibility rules.

Do you run this practice? Claim it — verified implementers get a public contact pathway and can propose corrections.

Data sources

Where this practice's information was retrieved from, and when.

Attachments

Similar practices you may find useful