# ChatGPT for Financial Services

> Source: https://aiwiki.ai/wiki/chatgpt_financial_services
> Updated: 2026-09-11
> Fact-checked: 2026-09-11
> Categories: AI Tools & Products, ChatGPT, Enterprise AI, OpenAI
> License: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/) - attribute to "AI Wiki (aiwiki.ai)"
> Cite as: AI Wiki. "ChatGPT for Financial Services." aiwiki.ai, 11 Sept 2026. https://aiwiki.ai/wiki/chatgpt_financial_services
> From AI Wiki (https://aiwiki.ai), the free encyclopedia of artificial intelligence. Reuse freely with attribution.

ChatGPT for Financial Services is an industry-specific product from [OpenAI](https://aiwiki.ai/wiki/openai), announced on September 10, 2026. OpenAI describes it as "a tailored ChatGPT Work experience that combines built-in financial data with GPT-6 Astra's reasoning to help teams develop research, financial models, and customized client materials".[1] It is a configuration of [ChatGPT Work](https://aiwiki.ai/wiki/chatgpt_work) rather than a new model: what it adds is a set of financial datasets that OpenAI licenses, indexes, and hosts itself, connectors tuned for market-data vendors, and a template mechanism that lets firm administrators fix the format of the Excel, Word, and PowerPoint files the product produces. OpenAI says the product was shaped by design partnerships with Morgan Stanley and Evercore, and that those partnerships steered where it started: investment banking and equity research.[1]

The announcement appeared on OpenAI's website and on the company's account on X the same day.[1][2] OpenAI says the product is "available to eligible financial institutions" and directs prospective customers to its sales team.[1]

## Key facts

| Field | Detail |
| --- | --- |
| Developer | OpenAI[1] |
| Announced | September 10, 2026[1][2] |
| Product type | Tailored [ChatGPT Work](https://aiwiki.ai/wiki/chatgpt_work) experience for financial institutions[1][2] |
| Headline model | [GPT-6 Astra](https://aiwiki.ai/wiki/gpt_6_astra), with newer models to be added "out of the box as they are released"[1] |
| Initial scope | Investment banking and equity research[1] |
| Design partners | Morgan Stanley, Evercore[1] |
| Built-in data providers | Daloopa, PitchBook, LSEG News; OpenAI's wording is "providers like" these three, so the list is illustrative[1] |
| Entitlement integrations in progress | S&P Capital IQ, LSEG, MSCI, Dow Jones Factiva, Moody's[1] |
| Connectors | "over 50+ connectors", with S&P Global and FactSet called out as optimized[1] |
| Security baseline | [ChatGPT Enterprise](https://aiwiki.ai/wiki/chatgpt_enterprise) SAML SSO, SCIM provisioning, role-based access control[1] |
| Availability | "available to eligible financial institutions"; contact sales[1] |

## Background

OpenAI was already selling into the sector before the product existed. Its Academy site carries a "Solution Kit for Financial Services" resource page, first published on December 10, 2025 and last updated on August 27, 2026, which offers prompt packs for [ChatGPT Enterprise](https://aiwiki.ai/wiki/chatgpt_enterprise), three example GPTs (a KYC and AML risk screener, a policy interpreter, and an investment research assistant), webinars, and whitepapers, and which points readers to a solutions page it calls "OpenAI for Financial Services".[5]

## Relationship to ChatGPT Work and ChatGPT Enterprise

ChatGPT Work is the agentic mode OpenAI introduced on July 9, 2026 for longer, multi-step projects that end in a finished artifact such as a document, spreadsheet, presentation, or interactive site. ChatGPT for Financial Services is presented as a version of that mode rather than a separate application: OpenAI's own phrasing is "a tailored ChatGPT Work experience".[1] The security model is inherited in the same way. OpenAI says the product "builds on ChatGPT Enterprise's SAML SSO, SCIM provisioning, and role-based access controls", which places it inside the existing enterprise administration stack instead of alongside it.[1]

What the vertical adds on top of Work is a data layer that OpenAI operates itself, connectors that OpenAI has tuned against financial-data vendors, a firm-template mechanism for output formatting, and a workspace arrangement for enforcing information barriers.[1]

## Design partners

OpenAI attributes the product's starting scope to work with two firms. "Our early work with Morgan Stanley and Evercore has helped steer where we have started: investment banking and equity research," the post says. "Reliable access to data and high quality artifact creation proved to be the biggest pain points for their teams." OpenAI adds that work with partners "will inform post training, product improvements, and our expansion into other financial services categories".[1]

Both firms supplied institutional statements rather than statements attributed to a named executive. Morgan Stanley's reads in part: "The promise of frontier research becomes real when it helps our people do the work that matters for our clients. We're working alongside OpenAI to bring that intelligence into how we research companies, develop analysis, and prepare advice."[1] Evercore's reads: "At Evercore, we focus on the questions that matter most to our clients, and we have an opportunity to apply frontier intelligence to help address them. We are working closely with OpenAI to shape how its technology can deepen the insights behind our advice, while building on the judgment and rigorous standards our clients expect from Evercore."[1]

Morgan Stanley's relationship with OpenAI predates this product by more than three years, though in a different division. On March 14, 2023 Morgan Stanley Wealth Management announced a "strategic initiative to create a bespoke solution with OpenAI", called itself "one of a handful of GPT-4 launch organizations" and "currently the only strategic client in wealth management receiving early access to OpenAI's new products", and described an internal service answering financial advisers' questions from the division's own research library, with responses "generated exclusively from MSWM content and with links to the source documents".[14] A later release from the same division says it "announced its relationship with OpenAI as its only wealth management strategic partner in March 2023 and fully rolled out the AI @ Morgan Stanley Assistant in September 2023", and that "To date, 98% of Financial Advisor teams have adopted the Assistant". That release, dated June 26, 2024, introduced AI @ Morgan Stanley Debrief, "an OpenAI-powered tool that, with client consent, generates notes on a Financial Advisors' behalf in client meetings and surfaces action items".[15] Both are wealth-management tools for advisers. The 2026 product is aimed at investment banking and equity research instead.

## The data layer

OpenAI frames data access, not model quality, as the problem the product is built around. The section heading is "Data is the foundation of all financial analysis".[1] The post describes three distinct mechanisms.

### Built-in premium data

The first is a set of datasets licensed by OpenAI and served from OpenAI's own infrastructure. The three named providers cover different ground: Daloopa normalizes company fundamentals from filings, PitchBook is Morningstar's private-markets data business and one of its five reportable segments,[18] and LSEG holds the exclusive right to distribute Reuters news to financial professionals.[16] The post describes the coverage this way: the product "includes datasets from providers like Daloopa, PitchBook, and LSEG News covering earnings transcripts, financial statements, company fundamentals, private companies, and more". The word "like" marks those three as examples rather than a complete list. OpenAI says teams can use these "immediately, with no separate contracts to negotiate or connectors to set up", and that "We index and host this data on OpenAI infrastructure, allowing us to improve retrieval, latency, and how we surface it in the product experience".[1]

Hosting the corpus is what OpenAI says makes its citation feature possible. The product offers "granular citations so that bankers can trace figures and claims back to their sources, and check the evidence as their analysis develops", with the underlying tables and passages highlighted for review. OpenAI's worked example is a profit-and-loss normalization: a banker can "inspect the reconciliation and notes behind an adjusted EBITDA, understanding which costs were excluded, and deciding how to use it in a valuation".[1] The company also says it will "post train our models to find, interpret, and use this data like we know the best analysts can", which is a statement of intent rather than a description of a shipped capability.[1]

| Built-in provider | Quoted in the announcement | Stated coverage |
| --- | --- | --- |
| Daloopa | Thomas Li, CEO | Named as a "native data partner"; fundamental financial data |
| PitchBook | Tom Van Buskirk, Executive Vice President of Technology and Engineering | Private-markets data |
| LSEG News | No quote in this section; LSEG is quoted under entitlement integrations below | News content. LSEG says it "is the exclusive distributor of Reuters news to the financial community under a 30-year news agreement that commenced in 2018"[16] |

Thomas Li's statement calls Daloopa "one of the native data partners for OpenAI for Financial Services" and says the deal "continues our strategy of being the data infrastructure for AI and agentic workflows in financial services".[1] Tom Van Buskirk's statement argues that "Partnering with OpenAI is a sign of where this industry is headed: toward answers built on trusted data, not just speed".[1] Li's wording names the OpenAI for Financial Services solutions programme rather than the product.[5]

Daloopa published its own announcement the same day, describing what it supplies more narrowly than the OpenAI post does: "Daloopa's select fundamental data is now natively available in ChatGPT for Financial Services", giving users "core financials and select KPIs, all verified and source-linked", and removing "the need for a user to toggle on connectors". That post carries an OpenAI quote that does not appear in OpenAI's own announcement, from Nick Turley, the company's vice president of product: "We want financial professionals to be able to work with the data they rely on, right inside ChatGPT. We're bringing together GPT-6 Astra and Daloopa's data in ChatGPT for Financial Services so teams can find the information they need, analyze it, and cite their sources in one place."[13]

Press accounts of the built-in tier are wider than the post. Reuters wrote that "The product includes datasets from Daloopa, LSEG News, PitchBook, Crunchbase, Quartr and others", CNBC described "native data access from LSEG, Daloopa, Crunchbase and PitchBook", and American Banker named PitchBook and Crunchbase.[8][9][10] Crunchbase and Quartr are not in the text of OpenAI's announcement as archived on the day of release: the post names Daloopa, PitchBook, and LSEG News, and Quartr appears only as a label in one chart (see Connectors below).[1] Where the reporters took the additional names from, whether a press briefing, a press kit, or a later edit to the page, is not recoverable from the sources here.

### Entitlement integrations

The second mechanism covers data a firm already licenses. OpenAI says it is "working with S&P Capital IQ, LSEG, MSCI, Dow Jones Factiva, and Moody's on shared sign-in and entitlement integrations", so that "Providers will be able to recognize users through their ChatGPT sign-in to enable automatic access to data they're already entitled to".[1] The tense here is forward-looking in the post; OpenAI describes the work as under way rather than complete.

| Subscription provider | Quoted in the announcement |
| --- | --- |
| S&P Capital IQ | Sally Moore, Chief Client Officer and Co-Head of Market Intelligence, Kensho Data & Platforms at S&P Global |
| LSEG | Emily Prince, Group Head of Enterprise AI, LSEG |
| MSCI | No quote |
| Dow Jones Factiva | No quote |
| Moody's | No quote |

Sally Moore's statement says that "Bringing the S&P Global AI Data Portal to ChatGPT gives financial professionals the full breadth of our verified intelligence, from financials and transcripts to market and energy data", and that S&P Global is collaborating with OpenAI "to deliver Adaptive and Deterministic Retrieval through ChatGPT for Financial Services".[1] Moore's reference to "Adaptive and Deterministic Retrieval" points at an existing S&P Global product rather than something built for OpenAI. On July 21, 2026 S&P Global announced Adaptive Retrieval, which lets customer agents and language models "access and assemble licensed S&P Global data using natural language queries", and said it would be sold together with the existing Deterministic Retrieval as "a single solution called the S&P Global AI Data Portal". Deterministic Retrieval is built on the Kensho LLM-ready API, which S&P Global says has been available to customers since 2025.[17] Emily Prince's statement ties the work to LSEG's own programme: "Our collaboration with OpenAI advances our LSEG Everywhere AI strategy. With MCP connectivity and a curated Reuters news selection, we're focused on making LSEG's trusted, licensed content and analytics available to customers wherever they want to work."[1]

### Connectors

The third mechanism is the ordinary [Model Context Protocol](https://aiwiki.ai/wiki/model_context_protocol) connector ecosystem, which OpenAI acknowledges has been awkward in this sector: "It can be challenging to effectively use MCP connectors for data providers." The company says it has optimized "some of the most used MCPs in financial services like S&P Global and FactSet" for immediate use, and that the broader ecosystem includes "over 50+ connectors including Datasite, Box, Preqin, and Intapp".[1]

OpenAI supports the claim with a chart titled "MCP Connector Quality Improvements", described in the page data as "Before and after error rates for financial services MCP connectors", attributing the improvement to "automated evaluation and iteration".[1] The figures are OpenAI's own, and the post does not define the error metric or the task set behind it.

| Connector | Error rate before | Error rate after |
| --- | --- | --- |
| Quartr | 5.09% | 1.99% |
| S&P Global | 6.84% | 2.66% |
| FactSet | 9.59% | 6.45% |
| Daloopa | 7.53% | 2.57% |

Quartr appears in this chart but is not named in the post's prose.[1]

## Model layer and benchmarks

OpenAI's claim for the model tier is that "GPT-6 Astra is state of the art across three of the core capabilities required for financial services work: information retrieval, financial reasoning, and artifact generation".[1] [GPT-6 Astra](https://aiwiki.ai/wiki/gpt_6_astra) began rolling out on September 3, 2026 and is OpenAI's current flagship. The post also says newer models will continue to arrive in the product "out of the box as they are released".[1]

Three evaluations are shown. All three are OpenAI-reported numbers about the model, not measurements of the financial-services product, and the comparison sets are chosen by OpenAI.

| Evaluation | Metric | GPT-6 Astra | GPT-5.6 Sol | Claude Fable 5.1 | Source of the eval |
| --- | --- | --- | --- | --- | --- |
| OfficeQA Pro | Correctness | 69.9% | 60.2% | 62.4% | Public benchmark published by Databricks[3] |
| BoxBench | Weighted rubric accuracy | 0.77 | 0.74 | 0.72 | "an Eval provided by an External partner", per OpenAI's footnote[1] |
| Professional slide creation | Win rate in human comparisons against [Claude Opus 5](https://aiwiki.ai/wiki/claude_opus_5) | 55.6% | 30.6% in the chart, 21.6% in the footnote (see below) | not shown | "This internal evaluation", per OpenAI's footnote[1] |

### OfficeQA Pro

Of the three, OfficeQA Pro is the only one with a methodology published outside OpenAI. It comes from [Databricks](https://aiwiki.ai/wiki/databricks) and was introduced in a paper submitted to arXiv on March 9, 2026 by a team including the company's chief technology officer Matei Zaharia. The benchmark asks agents 133 questions requiring parsing, retrieval, and analysis across a corpus of United States Treasury Bulletins spanning nearly 100 years, about 89,000 pages and more than 26 million numerical values. The paper reports that frontier models scored under 5% from parametric knowledge alone, under 12% with web access, and 34.1% on average when handed the corpus directly.[3] OpenAI's footnote describes the benchmark as testing "whether AI agents can find and analyze information across U.S. Treasury Bulletins, including complex financial tables, charts, and supporting footnotes", which matches the paper. The footnote names only GPT-5.6 Sol as a comparison, while the chart behind it also carries a Claude Fable 5.1 result that is higher than Sol's.[1][3]

Cross-vendor comparison on this benchmark is unreliable, and Anthropic says so directly. Its Claude Fable 5.1 and Claude Mythos 5.1 system card of September 1, 2026 reports 69.0% for [Claude Fable 5.1](https://aiwiki.ai/wiki/claude_fable_5_1) on OfficeQA Pro, against the 62.4% OpenAI's chart gives the same model. Anthropic calls OfficeQA "a public benchmark from Databricks" and OfficeQA Pro "a harder, 133-question subset of OfficeQA recommended for frontier models", says it evaluated agentically "with documents provided as extracted text in a sandboxed environment and with code-execution tools available", and adds a warning: "OfficeQA scores are highly sensitive to the evaluation harness. Settings that require the model to parse the raw PDF corpus directly yield substantially lower absolute scores for all models." As its illustration, Anthropic cites Databricks's own run of Claude Fable 5 reading the documents as images rather than extracted text, which scored 57.9% on OfficeQA Pro.[4] Neither OpenAI's post nor its footnote states which harness produced its figures.[1]

### BoxBench and the slide evaluation

For BoxBench, OpenAI's footnote says the benchmark "tests how well models can reason over documents in complex business workflows to turn them into rich deliverables" and adds: "Note this is an Eval provided by an External partner." The post does not name that partner, link a methodology, or say who ran the scoring.[1]

The slide-creation figure carries an internal contradiction in OpenAI's own materials. The footnote reads: "This internal evaluation tests models on professional slide creation, including both making slide decks from scratch and following existing templates. GPT-6 Astra scores a win rate of 55.6% against Opus 5 in human comparisons, compared to only 21.6% for GPT-5.6 Sol." The underlying chart data, however, labels 21.6% as GPT-5.5 and gives GPT-5.6 Sol 30.6%. Both the earliest and latest archived captures of the page carry the same conflict, so it is not an artefact of a later edit.[1]

## Artifacts and templates

Output formatting is handled through an administrator-published template library. OpenAI says that "administrators can publish Excel, Word, and PowerPoint templates through a dedicated admin page", and that with firm templates and style guides configured, teams "can turn their analysis into valuation models, research notes, and pitchbooks in their firm's format and style".[1] The product also builds interactive charts and visualizations with the underlying data and sources available for inspection.[1]

The announcement's own list of illustrated workflows is: value analysis, LBO modelling, buyer screening, earnings analysis, and pitchbook preparation.[1]

## Security and governance

OpenAI opens the governance section by naming the two constraints it is designing against: "Protecting material non-public information and client confidentiality is critical for financial institutions."[1] The stated controls are:

| Control | What OpenAI says |
| --- | --- |
| Identity | SAML single sign-on and SCIM provisioning, inherited from ChatGPT Enterprise[1] |
| Access | Role-based access controls; access to skills and apps manageable by role; supported app read and write actions can be enabled or disabled[1] |
| Training | "Your firm's business data is not used to train our models by default."[1] |
| Encryption | Encrypted at rest and in transit[1] |
| Retention | Administrators can configure workspace retention[1] |
| Audit | Compliance teams can export supported workspace logs through the OpenAI Compliance Platform into existing audit and investigation workflows[1] |
| Information barriers | "Multiple workspaces can also be created to enforce information barriers."[1] |

The information-barrier control works by separation rather than partition: a firm that needs a wall between two groups runs them in different workspaces rather than dividing one.[1]

## Availability

OpenAI states that "ChatGPT for Financial Services is available to eligible financial institutions" and asks interested firms to contact OpenAI or their account team. The post does not define eligibility, and its call to action is a sales contact rather than a self-service sign-up.[1]

Eligibility and price are the two things the announcement leaves open, and reporters pressed on both. American Banker said the company "did not immediately respond to a request for comment on what determines eligibility".[10] The Next Web reported that OpenAI "has not published a price", and TechRepublic that "OpenAI has not publicly disclosed pricing or minimum seat requirements".[11][12] Reuters reported that OpenAI "plans to expand the product beyond investment banking and equity research across the wider financial services sector", and CNBC that Turley said OpenAI intended tailored products for "a number of sectors" beyond finance.[8][9]

## Comparison with other financial-services AI products

The closest comparison is [Claude for Financial Services](https://aiwiki.ai/wiki/claude_for_financial_services), which [Anthropic](https://aiwiki.ai/wiki/anthropic) announced on July 15, 2025 and expanded on October 27, 2025. It is built on the same underlying idea: models bundled with prebuilt connectors to financial-data vendors plus implementation support. The partner rosters overlap heavily. Daloopa, PitchBook, S&P Global, FactSet, and Box are named in Anthropic's July 2025 announcement, and LSEG and Moody's were added in the October 2025 one; all seven appear in some form in OpenAI's post as well.[1][6][7]

The architectures differ in where the data sits. Anthropic's approach routes through MCP connectors to the vendor, so the vendor keeps and serves its own data. OpenAI does that too, but adds a first tier in which it licenses, indexes, and hosts the corpus itself, which is what its granular-citation feature depends on.[1] OpenAI's post presents the connector route as the harder path and the hosted route as the improvement: built-in data, it says, "remove the challenges with MCP connectors and access to data".[1]

OpenAI also says explicitly that the product is not intended to be its only answer for the sector: "ChatGPT for Financial Services is one way we serve customers across the industry, but we recognize that it will require a range of solutions to address the needs of the finance industry," with API-built applications named as the other route.[1]

The vertical pattern was not new in September 2026. Google Cloud announced Gemini Enterprise for Financial Services on August 25, 2026, in preview, in a post by chief executive Thomas Kurian that used the same design-partner framing: it named Deutsche Bank and CME Group as collaborators, with Deutsche Bank quoted as "a design partner for the Financial Research agent". Its architecture is described in the same terms as OpenAI's, built on prebuilt financial skills and "Secure Model Context Protocol (MCP) connectors", and its named data partners include LSEG, S&P Global, Moody's, and MSCI, four of the five firms in OpenAI's entitlement tier.[19] Microsoft took a horizontal route instead, adding what it calls federated Copilot connectors into Excel, Copilot Chat, and its Researcher agent. On June 25, 2026 it said connectors from LSEG and Moody's had been announced already and that CB Insights, Daloopa, FactSet (in preview), Morningstar, PitchBook, and S&P Global were being added.[20] The data vendors have also built their own tools: FactSet launched Pitch Creator on January 15, 2025, a pitchbook-automation product whose own subheading promised to "save junior bankers hours each week" and which draws on FactSet Mercury, the company's generative-AI chatbot.[21]

| Offering | Vendor | Announced | Shape |
| --- | --- | --- | --- |
| Claude for Financial Services | Anthropic | July 15, 2025, expanded October 27, 2025 | Model bundle plus prebuilt MCP connectors to data vendors and implementation support[6][7] |
| Gemini Enterprise for Financial Services | Google Cloud | August 25, 2026 (preview) | Vertical configuration of Gemini Enterprise with financial skills and MCP connectors, design partners Deutsche Bank and CME Group[19] |
| Federated Copilot connectors | Microsoft | June 25, 2026 for the finance batch | Horizontal: vendor data pulled into Excel, Copilot Chat, and the Researcher agent[20] |
| Pitch Creator | FactSet | January 15, 2025 | Single-workflow product for pitchbook creation, built on FactSet Mercury[21] |
| ChatGPT for Financial Services | OpenAI | September 10, 2026 | Vertical configuration of ChatGPT Work with OpenAI-hosted licensed data, tuned connectors, and firm templates[1] |

The same small set of data vendors appears on every one of these lists. What separates OpenAI's product from the others is the tier where it licenses and hosts the data itself instead of calling out to the vendor, which is the tier its granular citations rest on.[1]

## Reception

Coverage on September 10 and 11 turned mostly on one question: what the product does to the job of the junior banker. CNBC, which attended OpenAI's press briefing, ran the headline "OpenAI targets work of Wall Street junior bankers with new ChatGPT for Financial Services" and framed the launch as pushing OpenAI "deeper into territory traditionally occupied by Wall Street's entry-level bankers".[8] Reuters filed a shorter wire story built largely from the announcement.[9]

Nick Turley was OpenAI's spokesman at the briefing. "We're effectively teaching ChatGPT to research like an analyst and back up its conclusions like an analyst as well," he said. Demonstrating the product on a potential acquisition target, he argued that the hard part is judgement rather than formatting: "It's very easy to make slides that look good, but it's much harder to make slides [that] actually make sense. To get here, ChatGPT had to choose the relevant peers. It had to pull the prices into a spreadsheet. It had to check the chart against the data, and it had to explain the sell-off and the rebound." Asked directly by CNBC whether the product would reduce the need to hire junior bankers, he answered with an efficiency argument: "If you study the life of an analyst or of a banker, depending on the industry, they're working 100-hour weeks. I think in the same way that Microsoft Excel transformed the industry and allowed them to produce better analysis faster, you will see technology like this do the same." Turley said there was "a ton of demand" but declined to name any bank that had signed on.[8]

CNBC also placed the launch in OpenAI's commercial context, noting that chief financial officer Sarah Friar told investors in August that the enterprise business accounted for more revenue than the consumer business, and that the launch came as the company prepared for an expected initial public offering.[8] American Banker reported that the announcement followed OpenAI's confidential preliminary filings to the SEC and its launch of a Plaid integration for individual ChatGPT Pro users.[10]

American Banker quoted two outside experts, who differed on how far the substitution goes. Annie DeStefano, a fintech and banking consultant formerly at Goldman Sachs and Silicon Valley Bank, told American Banker that "The depth [that] regulated industries require for AI adoption will absolutely require this type of industry-led design partnership", and that "When dealing with sensitive information and transactions, continued judgement and discernment by trained professionals, within an established risk framework of an institution, will be needed". Theodora Lau, founder of Unconventional Ventures, put the objection in terms of training rather than accuracy: "We are removing the ability for the junior bankers to learn what a wrong model looks like. Ten years down the road, you'll end up with managing directors who have never built a model from scratch. That's what apprenticeship is supposed to be for, and it just got automated by OpenAI."[10] CNBC quoted a related warning made a month earlier by Chris Churchman, the Goldman Sachs partner running one of the bank's flagship AI projects, about "cognitive atrophy" in the next generation of financiers.[8]

The Next Web raised a regulatory gap. The announcement, it observed, "does not mention EU data residency, a specific European hosting region, or the Digital Operational Resilience Act, known as DORA", rules that have governed third-party information and communications technology arrangements for EU financial firms since January 2025. The same piece noted that Morgan Stanley, one of the two design partners, employs analysts who have forecast that European banks could lose a fifth of their jobs to AI.[11]

Several outlets flattened OpenAI's three data tiers into one list, which matters because the tiers differ in who holds the data and who pays for it. Reuters, for instance, placed Preqin and Datasite in the entitlement tier, where OpenAI's post puts them in the wider connector ecosystem.[1][9] TechRepublic kept the tiers apart.[12]

## References

1. OpenAI. "Introducing ChatGPT for Financial Services." September 10, 2026. https://openai.com/index/introducing-chatgpt-financial-services/ (verified against Internet Archive captures https://web.archive.org/web/20260910175211/https://openai.com/index/introducing-chatgpt-financial-services/ and https://web.archive.org/web/20260910185345/https://openai.com/index/introducing-chatgpt-financial-services/; benchmark chart values and footnote text taken from the page's embedded data)
2. OpenAI. "Now available: ChatGPT for Financial Services." Official OpenAI account on X, September 10, 2026. https://x.com/OpenAI/status/2098118191029624911
3. Krista Opsahl-Ong, Arnav Singhvi, Jasmine Collins, Ivan Zhou, Cindy Wang, Ashutosh Baheti, Owen Oertell, Jacob Portes, Sam Havens, Erich Elsen, Michael Bendersky, Matei Zaharia, Xing Chen. "OfficeQA Pro: An Enterprise Benchmark for End-to-End Grounded Reasoning." arXiv:2603.08655, March 9, 2026. https://arxiv.org/abs/2603.08655
4. Anthropic. "Claude Fable 5.1 & Claude Mythos 5.1 System Card." September 1, 2026, section 8.15.1. https://www-cdn.anthropic.com/0339e6a7c5c7b87f5c07798616dc32c215d14235/Claude%20Fable%205.1%20&%20Claude%20Mythos%205.1%20System%20Card.pdf
5. OpenAI Academy. "ChatGPT for financial services." Published December 10, 2025, last updated August 27, 2026. https://academy.openai.com/public/clubs/work-users-ynjqu/resources/finserv
6. Anthropic. "Claude for Financial Services." July 15, 2025. https://www.anthropic.com/news/claude-for-financial-services
7. Anthropic. "Advancing Claude for Financial Services." October 27, 2025. https://www.anthropic.com/news/advancing-claude-for-financial-services
8. Hugh Son and Ashley Capoot. "OpenAI targets work of Wall Street junior bankers with new ChatGPT for Financial Services." CNBC, September 10, 2026. https://www.cnbc.com/2026/09/10/openai-chatgpt-for-financial-services-targets-work-of-junior-bankers.html
9. Akash Sriram and Krystal Hu. "OpenAI launches ChatGPT for financial services industry." Reuters, September 10, 2026 (read via AOL syndication). https://www.aol.com/articles/openai-launches-chatgpt-financial-services-172936000.html
10. Melinda Lucy. "OpenAI launches financial tool for Wall Street bankers." American Banker, September 10, 2026. https://www.americanbanker.com/news/openai-launches-financial-tool-for-wall-street-bankers
11. "OpenAI launches ChatGPT for Financial Services with data from S&P, LSEG, Moody's and PitchBook." The Next Web, September 11, 2026. https://thenextweb.com/news/openai-chatgpt-financial-services-dora-data-residency
12. Aminu Abdullahi. "OpenAI Launches ChatGPT for Financial Services: What Banks Should Know." TechRepublic, September 11, 2026. https://www.techrepublic.com/article/news-openai-chatgpt-financial-services-banks/
13. Daloopa. "Daloopa Brings Verified Financial Data Directly into ChatGPT for Financial Services." September 10, 2026. https://daloopa.com/blog/product-updates/announcing-daloopa-partnership-with-openai
14. Morgan Stanley. "Morgan Stanley Wealth Management Announces Key Milestone in Innovation Journey with OpenAI." March 14, 2023. https://www.morganstanley.com/press-releases/key-milestone-in-innovation-journey-with-openai
15. Morgan Stanley. "Morgan Stanley Wealth Management Announces Latest Game-Changing Addition to Suite of GenAI Tools." June 26, 2024. https://www.morganstanley.com/press-releases/ai-at-morgan-stanley-debrief-launch
16. LSEG. "LSEG and Thomson Reuters announce joint commitments to enhance the value of financial and markets news service as part of long-term partnership." January 17, 2023. https://www.lseg.com/en/media-centre/press-releases/2023/lseg-and-thomson-reuters-enhance-value-financial-markets-news-service
17. S&P Global. "S&P Global launches Adaptive Retrieval, giving customers a new way to access data across AI and agentic workflows." July 21, 2026 (published on kensho.com). https://kensho.com/news/sp-global-launches-adaptive-retrieval-and-ai-data-portal-enabling-data-across-ai-agentic-workflows
18. Morningstar, Inc. Form 10-K for the fiscal year ended December 31, 2025, filed February 13, 2026. https://www.sec.gov/Archives/edgar/data/1289419/000128941926000013/morn-20251231.htm
19. Thomas Kurian. "Now introducing Gemini Enterprise for Financial Services." Google Cloud Blog, August 25, 2026. https://cloud.google.com/blog/products/ai-machine-learning/introducing-gemini-enterprise-for-financial-services
20. Bill Borden. "AI in financial services: Bringing trusted data into the flow of work." Microsoft Cloud Blog, June 25, 2026. https://www.microsoft.com/en-us/microsoft-cloud/blog/financial-services/2026/06/25/ai-in-financial-services-bringing-trusted-data-into-the-flow-of-work/
21. FactSet. "FactSet Launches AI-Powered Pitch Creator." January 15, 2025. https://investor.factset.com/news-releases/news-release-details/factset-launches-ai-powered-pitch-creator