How does Standard Chartered govern AI?
Standard Chartered has had a principles-based responsible AI standard since 2021, run through a Responsible AI Council that assesses each model and decides whether to approve it. A central AI Safety Council of engineering, risk, compliance and business leaders oversees the bank's AI inventory, and the 2025 annual report says the framework aligns with MAS FEAT and the HKMA big data and AI guidelines, with the Audit Committee receiving twice-yearly reports on data risk including responsible AI. Alvaro Garrido says new models are evaluated against three impact dimensions, loss of data, loss of funds or loss of service, with preventive and detective controls setting acceptable residual risk. At a 22 September roundtable reported on 28 October 2026 Group Chief Data Officer Shebani Baweja said 80% of key data controls are codified and 90% continuously monitored through automated AI guardrails, and that the bank is moving data controls into design and development. On agents, the bank keeps humans accountable: its markets CIO said in November 2026 that agentic AI "supports decision-making, it doesn't do decision-making". The half-year 2026 risk review lists rapid AI adoption among topical risks.[1][2][3][4][5][6]
Data as of Nov 17, 2026 · Standard Chartered profile updated Nov 17, 2026 · 20 sources cited on this page · 141 banks profiled on this site · Markdown
Which bodies and standards govern AI at Standard Chartered?
| Item | What it does | Source date |
|---|---|---|
| Responsible AI standard | Principles-based standard in place since 2021 | 24 July 2025[1] |
| Responsible AI Council | Assesses each model against fairness, ethics, transparency, performance, cyber and privacy principles and decides whether to approve it; chaired in July 2025 by the then group chief data officer | 24 July 2025[1] |
| AI Safety Council | Central council of engineering, risk, compliance and business leaders with cross-functional oversight of the AI inventory, so deployments meet consistent standards globally | Undated (The Asian Banker)[2][3] |
| Impact dimensions | New models are evaluated against loss of data, loss of funds or loss of service, with preventive and detective controls defining acceptable residual risk | Undated (The Asian Banker)[2] |
| 360-degree framework | Brings together security, third-party risk and AI assessments in one review | Undated (The Asian Banker)[2] |
| Audit Committee | Receives twice-yearly reports on data risk, including responsible AI; framework aligned with MAS FEAT and the HKMA big data and AI guidelines | 24 February 2026[3] |
| Data controls | 80% of key data controls codified and 90% continuously monitored through automated AI guardrails; 95% data accuracy reported in AI document processing, scope not specified | 22 September (reported 28 October 2026)[4] |
| AI safety | David Hardoon, then global head of AI enablement, preferred the term "AI safety" over "responsible AI" and built on the FEAT principles he helped pioneer at MAS | Undated (The Asian Banker)[7] |
How does Standard Chartered keep humans in control of AI agents?
Gareth Hughes, CIO for markets, said in November 2026 that AI summarising documents with citations needs less scrutiny than systems recommending business decisions, and that the bank's approach to agentic AI keeps human oversight: "It supports decision-making, it doesn't do decision-making." Engineers review AI outputs at each stage of specification-driven development.[5]
In transaction banking, Global Head of Digital Channel Solutions Sunday Domingo described agentic payments as a progression from visibility to recommendation to delegation, and said agents would get permissions and guardrails built on controls that already exist in the Straight2Bank platform, where corporate users and authorisers are onboarded and assigned permissions (October 2026).[8]
The bank's April 2026 paper on agentic AI in asset servicing describes a client service manager who stays "human in the loop", reviewing output from AI operator agents before it reaches the client. Agent-enabled employee onboarding likewise leaves the final risk decision with humans, and at Trust Bank the incident agents recommend rather than act.[9][10][11]
What has Standard Chartered not published about AI governance?
The bank has not published its responsible AI standard, the membership of either council or the number of models each has approved or rejected. The AI Safety Council and impact-dimension description come from executive interviews and the annual report, not a standalone policy document.[1][2][3]
The half-year 2026 risk review names AI-enabled cyber threats and frontier AI (Mythos) and lists rapid AI adoption among topical risks, but gives no AI incident figures. Hong Kong and Singapore regulators sought clarification of the CEO's May 2026 AI-and-jobs remarks, according to Bloomberg via Reuters; no supervisory finding has been published.[6][12]
What does the Standard Chartered profile record on this topic?
The rows below are the entries of the StanChart AI strategy profile that bear on this page, as of Nov 17, 2026; the profile holds the full record of 37 dated moves, 21 use cases and 65 sources.
| Use case | What the record shows | Status | Source |
|---|---|---|---|
| AI Safety Council and tiered evaluation of AI systems | The 2025 annual report says the council evaluates AI solutions before use, systems are assessed in proportion to their risk, and the Audit Committee receives twice-yearly reports on data risk including responsible AI, aligned with MAS FEAT and the HKMA big data and AI guidelines. | In production | [3] |
| Responsible AI standard and Responsible AI Council | A principles-based standard in place since 2021; the council assesses each model against fairness, ethics, transparency, performance, cyber and privacy principles and decides whether to approve it. | In production | [1] |
| Fairness metrics for AI and GenAI systems | Research with A*STAR IHPC that surveyed, established and qualified a range of fairness metrics so AI and GenAI solutions stay fair as data and societal values change. | Pilot | [13] |
| Date | Move | Detail | Source |
|---|---|---|---|
| Jul 29, 2026 | Half-year report: AI as opportunity and risk, Mythos named | The risk review signed by Group Chief Risk Officer Jason Forrester says AI enhances operational efficiency but raises AI-enabled cyber threats, cites frontier AI (Mythos), and lists rapid adoption of AI among topical risks covering governance, data privacy, resilience and workforce disruption. | [6] |
| Mar 20, 2026 | MAS AI risk toolkit lists Standard Chartered among 24 MindForge participants | The toolkit covers traditional, generative and agentic AI in three documents of 26, 173 and 23 pages; participants include DBS, Julius Baer, OCBC, Standard Chartered and UOB, and MAS will set up an AI risk management workgroup under BuildFin.ai. | [14] |
| Oct 15, 2025 | HKMA second GenA.I. Sandbox cohort again includes Standard Chartered Bank (Hong Kong) | 27 use cases from 20 banks and 14 technology partners were selected from more than 60 proposals, with trials starting in early 2026 on Cyberport’s AI Supercomputing Centre; the annex lists Standard Chartered Bank (Hong Kong) Limited. | [15][16] |
| Jul 24, 2025 | Responsible AI standard and council described | The group chief data officer says a principles-based standard in place since 2021 is operationalised through a Responsible AI Council he chairs, which assesses each model and decides whether to approve it. | [1] |
| Apr 23, 2025 | SC Ventures and A*STAR IHPC complete two GenAI research engagements | One project used large language models and retrieval-augmented generation to map regulatory obligations and analyse regulatory change; the other, with Standard Chartered, surveyed and qualified fairness metrics for AI and GenAI systems. | [13] |
| Dec 19, 2024 | HKMA names Standard Chartered Bank (Hong Kong) in the first GenA.I. Sandbox cohort | The HKMA and Cyberport selected 15 use cases from 10 banks and four technology partners from more than 40 proposals; Standard Chartered Bank (Hong Kong) Limited is one of the ten banks. The annex gives example themes (AI-assisted financing approval, AI-powered AML suspicious transaction reporting, an assistant for fraud investigators) rather than the bank’s own project. | [17][18] |
| Nov 15, 2023 | Member of MAS's Project MindForge generative AI consortium | MAS said phase one of Project MindForge produced a GenAI risk framework with seven risk dimensions; the consortium comprised DBS, OCBC, UOB, Standard Chartered Bank, Citi Singapore, HSBC, Google Cloud, Microsoft, MAS, Accenture and the Association of Banks in Singapore. | [19] |
| Jun 26, 2023 | One of seven banks piloting MAS's Veritas FEAT methodology | MAS's Veritas consortium released Toolkit 2.0 and a white paper on lessons from seven institutions, including Standard Chartered Bank, that piloted integrating the Veritas methodology with their internal governance frameworks. | [20] |
Does Standard Chartered have a responsible AI framework?
Yes. A principles-based responsible AI standard has been in place since 2021, operated through a Responsible AI Council that assesses each model and decides whether to approve it. The 2025 annual report says the framework aligns with MAS FEAT and the HKMA big data and AI guidelines.
What is Standard Chartered's AI Safety Council?
A central council of engineering, risk, compliance and business leaders that oversees the bank's AI inventory and evaluates AI solutions before use, so that deployments meet consistent standards across markets.
How does Standard Chartered assess AI risk?
According to Alvaro Garrido, new models are evaluated against three impact dimensions: loss of data, loss of funds or loss of service, with preventive and detective controls defining acceptable residual risk. Higher-risk applications get more rigorous review than lower-risk ones.
Does Standard Chartered let AI agents make decisions?
Not on its own account. Its markets CIO said in November 2026 that its agentic AI "supports decision-making, it doesn't do decision-making", and agent designs in onboarding, asset servicing and payments keep a human approver or defined permissions.
Who oversees AI risk at board level?
The 2025 annual report says the Audit Committee receives twice-yearly reports on data risk, including responsible AI. Group Chief Risk Officer Jason Forrester signed the half-year 2026 risk review that names AI adoption as a topical risk.
- Developing safe and human centric AI
- Standard Chartered moves into its next phase of disciplined AI execution (undated page, retrieved 2026-11-17)
- Annual Report 2025
- How is Standard Chartered building the controls needed to scale AI?
- Can Standard Chartered scale its markets technology while improving operating efficiency?
- Half Year Report 2026
- Standard Chartered’s risk-first approach to scaling AI globally (undated page, retrieved 2026-11-17)
- Can Standard Chartered move agentic payments from recommendation to delegated action?
- Agentic AI in Asset servicing: The operating model for scalable and trusted autonomy
- Transformation Transcript: May 2026 Investor Event
- Trust Bank cuts incident triage time to two minutes with AI agents
- StanChart CEO apologises for 'upset caused' by AI comments
- SC Ventures partners with A*STAR to leverage AI for smarter regulatory compliance
- MAS launches AI risk toolkit for financial institutions with case studies from DBS, peers
- HKMA announces second cohort of GenA.I. Sandbox
- Annex: Second cohort of GenA.I. Sandbox and examples of proposed use cases
- HKMA announces first cohort of GenA.I. Sandbox
- Annex: First cohort of GenA.I. Sandbox and examples of proposed use cases
- MAS Partners Industry to Develop Generative AI Risk Framework for the Financial Sector
- MAS-led Industry Consortium Releases Toolkit for Responsible Use of AI in the Financial Sector
When Standard Chartered moves on AI, you'll read it here first.
when one of the banks we cover moves on AI, it is in the next morning's brief · six sourced stories · 7 am PT · free
plus every regulation, bank and agent page update, the morning after · leave any morning