AI incidents · computed from AI incidents register and bank profiles; nothing on this page is written by a model. All ranked answers: /banks/leaders
On the public record as of Oct 28, 2026, 17 of the 141 large banks profiled have a dated record of restricting employee use of an AI tool or pausing an AI deployment. Ranked by records, the top 5 are Goldman (4 records), DBS (1 record), Associated (1 record), HSBC (1 record) and Commonwealth Bank (1 record). Goldman's strongest evidence: Goldman Sachs removes Anthropic Claude access for its Hong Kong bankers (llm restriction, Apr 28, 2026).
141 banks profiled (120 US, 21 non-US) · 17 with matching evidence · ranked answer, not a comparison
Which of the large banks have restricted or paused AI tools?
| # | Bank | Records | Latest record | Status | Date | On this site | Sources |
|---|---|---|---|---|---|---|---|
| 1 | Goldman | 4 | Goldman Sachs removes Anthropic Claude access for its Hong Kong bankers (llm restriction, Apr 28, 2026) | llm restriction | Apr 28, 2026 | Responses list → | Reuters ↗ · The Wall Street Journal ↗ |
| 2 | DBS | 1 | DBS holds off on letting AI agents act on their own while controls lag capability (paused deployment, Jul 28, 2026) | paused deployment | Jul 28, 2026 | Responses list → | Computer Weekly ↗ |
| 3 | Associated | 1 | Associated Bank removes its 'Adie' virtual assistant from its digital-banking terms (paused deployment, Jun 22, 2026) | paused deployment | Jun 22, 2026 | Responses list → | Associated Bank ↗ |
| 4 | HSBC | 1 | OpenAI opens GPT-5.5 Cyber to nine UK banks, including HSBC (timeline, Jun 1, 2026) | policy move | Jun 1, 2026 | HSBC profile → | BBC News ↗ |
| 5 | Commonwealth Bank | 1 | Banks test OpenAI's cyber model (timeline, May 19, 2026) | policy move | May 19, 2026 | Commonwealth Bank profile → | Australian Financial Review ↗ |
| 6 | MUFG | 1 | Reported access to Claude Mythos and an OpenAI cyber-defence model (timeline, May 13, 2026) | policy move | May 13, 2026 | MUFG profile → | Nikkei Asia ↗ |
| 7 | Synchrony | 1 | Illinois datathon uses Synchrony call-centre data (timeline, Mar 27, 2026) | policy move | Mar 27, 2026 | Synchrony profile → | University of Illinois Urbana-Champaign, Office of Data Science Research ↗ |
| 8 | ConnectOne | 1 | 10-K describes AI already in use and a ban on non-approved tools (timeline, Feb 24, 2026) | policy move | Feb 24, 2026 | ConnectOne profile → | ConnectOne Bancorp (SEC filing) ↗ |
| 9 | Société Générale | 1 | Société Générale scraps its in-house GenAI assistant SoGPT in favour of Microsoft Copilot (paused deployment, Jan 12, 2026) | paused deployment | Jan 12, 2026 | Responses list → | Bloomberg ↗ |
| 10 | UBS Group AG | 1 | UBS slows rollout of AI video avatars of its analysts after accent problems (paused deployment, May 19, 2025) | paused deployment | May 19, 2025 | Responses list → | Financial Times ↗ |
| 11 | Raymond James | 1 | CAIO interview: no in-house LLMs, vendors barred from training on firm data (timeline, Mar 10, 2025) | policy move | Mar 10, 2025 | Raymond James profile → | Financial Planning ↗ |
| 12 | First Horizon | 1 | First Horizon says AI 'burnout' resets for call-centre staff were not rolled out after backlash (paused deployment, Jul 3, 2024) | paused deployment | Jul 3, 2024 | Responses list → | American Banker ↗ |
| 13 | JPMorgan | 1 | JPMorgan, Bank of America, Citi, Goldman Sachs, Deutsche Bank and Wells Fargo restrict staff use of ChatGPT (llm restriction, Feb 22, 2023) | llm restriction | Feb 22, 2023 | Responses list → | Bloomberg ↗ |
| 14 | BofA | 1 | JPMorgan, Bank of America, Citi, Goldman Sachs, Deutsche Bank and Wells Fargo restrict staff use of ChatGPT (llm restriction, Feb 22, 2023) | llm restriction | Feb 22, 2023 | Responses list → | Bloomberg ↗ |
| 15 | Citi | 1 | JPMorgan, Bank of America, Citi, Goldman Sachs, Deutsche Bank and Wells Fargo restrict staff use of ChatGPT (llm restriction, Feb 22, 2023) | llm restriction | Feb 22, 2023 | Responses list → | Bloomberg ↗ |
| 16 | Wells Fargo | 1 | JPMorgan, Bank of America, Citi, Goldman Sachs, Deutsche Bank and Wells Fargo restrict staff use of ChatGPT (llm restriction, Feb 22, 2023) | llm restriction | Feb 22, 2023 | Responses list → | Bloomberg ↗ |
| 17 | Deutsche Bank | 1 | JPMorgan, Bank of America, Citi, Goldman Sachs, Deutsche Bank and Wells Fargo restrict staff use of ChatGPT (llm restriction, Feb 22, 2023) | llm restriction | Feb 22, 2023 | Responses list → | Bloomberg ↗ |
Each row quotes one record from AI incidents register and bank profiles; the “On this site” link opens the bank's own page or sub-page with every source. Banks absent from the table have nothing matching on the record this site holds.
How is this ranked?
Data: the AI incidents at banks register (/ai-incidents): dated, sourced records of AI failures, AI used against banks and governance findings, each with a status and severity, its responses list (/ai-incidents/responses), plus profile timeline moves that restrict or block an AI tool.
Rule: Responses of kind "LLM restriction" or "paused deployment" naming the bank, plus profile timeline moves whose text blocks, bars, restricts or bans an AI tool. Score = records; ties by the newest record.
Counts: 17 of 141 profiled banks qualify; 14 distinct sources cited in the table. Data checked 2026-10-28; ranking template updated 2026-10-30.
What are the caveats?
A restriction is a policy at a point in time; several banks later lifted theirs, and the newest record is shown.
Banks absent from the table have nothing matching on the record this site holds, which is not evidence that they do nothing. Each row links the bank's own page, where every entry carries its sources.
Which bank ranks first?
Goldman Sachs leads with 4 records: Goldman Sachs removes Anthropic Claude access for its Hong Kong bankers (llm restriction, Apr 28, 2026). DBS is second with 1.
How many large banks restricted or paused AI tools?
17 of the 141 profiled banks (120 US, 21 non-US groups). Among those listed, 10 are US banks and 7 non-US groups.
What counts, and what does not?
Responses of kind "LLM restriction" or "paused deployment" naming the bank, plus profile timeline moves whose text blocks, bars, restricts or bans an AI tool. Score = records; ties by the newest record. Source: the AI incidents at banks register (/ai-incidents): dated, sourced records of AI failures, AI used against banks and governance findings, each with a status and severity, its responses list (/ai-incidents/responses), plus profile timeline moves that restrict or block an AI tool.
Is this a measure of capability?
No. Every row is a sourced fact on the public record: what the bank, a regulator, a filing or an independent outlet has put on the record. The page does not rate how well anything works.
When was this checked, and how often does it update?
The data behind the table was last checked on 2026-10-28 (AI incidents register and bank profiles). The page recomputes whenever that dataset changes and is republished with the site; it is never edited by hand.
When any of these banks moves on AI, you'll read it the next morning.
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