A multi-model, multi-cloud AI bank: its own LLM-as-a-Service platform, Mistral AI as the European partner, Google Cloud for agentic AI, and AI written into the 2027–2030 cost plan.
How is BNP Paribas using AI?
BNP Paribas is the eurozone bank that most explicitly puts AI into its financial plan. Its second-quarter 2026 results slides cite the Evident AI index as ranking it first among eurozone banks, say AI is 'at the heart of' its 2027–2030 plan, and raise targeted annual cost savings from about €700 million (2022–2026 average) to about €1 billion (2027–2030 average). The technology is a multi-model stack: an internal LLM as a Service platform deployed in June 2025 in the Group's own data centres, a renewed three-year partnership with Mistral AI announced on 26 May 2026, and a five-year Google Cloud partnership for Gemini Enterprise and agentic AI announced on 24 September 2026, while LLM@CIB is already available to more than 65,000 employees of the corporate and investment bank. BNP Paribas says it runs more than 800 AI use cases, trained nearly 80,000 employees in AI in 2025, had 100,000 people onboarded to its 'My companion' Group AI platform by July 2026, and governs AI through a Group AI Committee, an AI Risk Committee and AI strategy committees chaired by the CEO.
Lead bank
BNP Paribas SA
Total assets (group, as reported, Dec 31, 2025)
€2,793.0bn total consolidated balance sheet[6]
Charter
French bank holding group (société anonyme); supervised by the ECB under the Single Supervisory Mechanism with the ACPR as French authority
Headquarters
Paris, France
Flagship platform
LLM as a Service, with Mistral AI and Google Cloud
AI leadership
Jean-Laurent Bonnafé · Marc Camus · Charles Holive · Sophie Heller · Olivier Osty
- AI is in the plan, not just the pilot list: the Q2 2026 slides say AI is 'at the heart of' the 2027–2030 plan, that the Group is 'moving from fragmented AI initiatives to industrialised execution at scale, paving the way to the Agentic era', and that AI benefits are expected to ramp up throughout the plan; Reuters reported in February 2026 that AI is expected to deliver around half of the next incremental efficiency gains.
- The CEO is selective about spend: Jean-Laurent Bonnafé told reporters on 5 February 2026 that the bank is investing where it makes sense because AI is really expensive, describing it as a tool to cut running costs and simplify procedures across support functions.
- Multi-model by design: LLM as a Service (June 2025) offers open-source and Mistral AI models inside the Group's data centres; Mistral AI, a partner since 2023 and Mistral AI's longest-standing financial-services partner, was renewed for three years on 26 May 2026; Google Cloud adds Gemini Enterprise and agents for corporate and investment banking from 24 September 2026, with some data categories not stored in the public cloud.
- Named use cases span the franchise: LLM@CIB for more than 65,000 CIB employees, HelloïZ for more than one million Hello bank! clients since January 2026, Nickel Assist for 200 customer advisers on Gemini, an ESG-assessment assistant used for around 35% of ESG assessments, generative-AI tools at the Inspection Générale, and planned AI agents preparing corporate credit memos.
- Governance is formal: the 2025 Universal Registration Document describes a Group AI Committee (GAIC) that approves the AI strategy and evaluates use cases against risk assessments, and an AI Risk Committee that monitors AI Act compliance and model attestation; the Q2 2026 slides add Group AI strategic committees chaired by the CEO.
- People investment is large: nearly 80,000 employees received AI-related training in 2025 and more than 30,000 have completed the Introduction to GenAI module; five responsible-AI principles (fairness, transparency, data and rights protection, human oversight, skills) apply, with a sixth on environmental footprint being introduced.
- Research is in-house and academic: the FinAI-LAB joint laboratory with Télécom Paris and a collaboration with Mila cover AI against financial crime, agentic systems and foundation models for finance, trustworthy AI and adaptive models.
- Cyber is part of the AI case: CIO Marc Camus said on 26 May 2026 that the speed and scale at which AI can find vulnerabilities is a fundamental shift, and the Mistral partnership extends to cyber defences and KYC on-premises solutions. BNP Paribas's US activities are not covered on this page.
What is LLM as a Service, with Mistral AI and Google Cloud?
BNP Paribas built an internal LLM as a Service platform, hosted in its own data centres on GPU infrastructure, that gives Group entities unified access to open-source models, Mistral AI models and, later, models trained on internal data (announced 5 June 2025). Around it sit a three-year renewal of the Mistral AI partnership (26 May 2026), a five-year Google Cloud partnership giving access to Gemini Enterprise and Gemini models for agentic AI (24 September 2026), LLM@CIB, the corporate and investment bank's assistant available to more than 65,000 employees, and a Group AI platform for staff branded 'My companion' that had onboarded 100,000 people by the second-quarter 2026 results. The bank describes the approach as multi-cloud and multi-model, with each AI agent authenticated and limited to the resources its task needs.[1][9][15][12][17]
What has BNP Paribas done on AI, and when?
- Sep 24, 2026
Five-year Google Cloud partnership for agentic AI[15][17]
BNP Paribas and Google Cloud announced a five-year partnership giving access to Gemini Enterprise and Gemini models, starting with agents for corporate credit memos in CIB and Gemini inside LLM@CIB; Nickel Assist already runs on Gemini for 200 customer advisers.
- Sep 22, 2026
Strategy update ties AI to support-function efficiency[14]
BNP Paribas said its support-functions transformation plan would further accelerate efficiency gains through technology simplification, process industrialisation and the large-scale deployment of AI, while preparing its 2030 strategic plan.
- Sep 15, 2026
Technology function commits to agentic AI investment in 2026[13]
In a statement written on 12 June 2026 and published on 15 September, the Group said it would invest in agentic AI in 2026 to meet business-line and function needs and optimise end-to-end processes, keeping service continuity, security and cost control as its three priorities, and cited the LLM as a Service platform as a foundation it had built in 2025.
- Jul 23, 2026
Q2 2026 results: AI at the heart of the 2027–2030 plan[12]
The results slides state that the Group is moving from fragmented AI initiatives to industrialised execution at scale, cite the Evident AI index ranking it first among eurozone banks, report 100,000 people onboarded to the 'My companion' Group AI platform and raise annual savings from about €700 million to about €1 billion.
- Jul 21, 2026
Responsible AI principles published[11]
BNP Paribas set out five guiding principles for AI (fairness and non-discrimination, transparency and explainability, protection of data and fundamental rights, human oversight, skills development), said a sixth pillar on environmental footprint is being introduced, and cited nearly 80,000 employees trained in AI in 2025.
- May 26, 2026
Mistral AI partnership renewed for three years[9][10]
BNP Paribas renewed its partnership with Mistral AI, extending it beyond model access to software, solutions and co-development research, with priority use cases in internal knowledge and research, on-premises KYC solutions and client and employee assistants; the CIO also said the bank was preparing for AI-driven vulnerability discovery.
- Mar 26, 2026
AI companion to be rolled out to all staff[7]
Bloomberg reported that BNP Paribas would deploy an AI companion for all staff, progressively, giving broader access to generative AI while keeping specialised assistants built in business units, as the CEO bets AI can lower costs.
- Mar 19, 2026
2025 Universal Registration Document published[6]
The annual report describes the Group AI Committee, the AI Risk Committee (AI Act compliance and model attestation), generative-AI audit tools at the Inspection Générale and AI-related training, and its Pillar 3 appendix reproduces the Group's global-risk landscape, which rates 'Risks related to artificial intelligence' High in the short term, Very High in the medium term and Extremely High in the long term.
- Feb 5, 2026
Full-year 2025 results: AI to help deliver cost cuts[5]
The CEO said AI would help deliver further savings by streamlining procedures and cutting running costs in support functions but would be deployed selectively; the bank said its next three-year plan, due in early 2027, would include a comprehensive review of processes with AI expected to deliver around half of the next incremental efficiency gains.
- Jan 28, 2026
Charles Holive named AI chief of the corporate and investment bank[4]
Banking Dive reported that BNP Paribas had recently named Charles Holive, formerly of JPMorgan Chase and PepsiCo, as its corporate and investment bank's AI chief.
- Nov 12, 2025
Generative AI assistant for ESG assessments[3]
BNP Paribas described an internal assistant, built on ChatGPT and Microsoft Azure, that proposes answers to the up-to-80-question ESG Assessment from client documents; 600 users had completed almost 1,600 assessments with it, around 35% of all ESG Assessments.
- Jul 8, 2025
GenAI fintech incubator for BNP Paribas Personal Finance UK[2]
BNP Paribas Personal Finance UK opened applications for an incubator for start-ups building generative-AI models for banks, with the best to be tested on the 100,000 customers of its myCreation app.
- Jun 5, 2025
LLM as a Service platform deployed[1][2]
BNP Paribas announced an internal LLM as a Service platform, operated by Group IT in its own GPU-equipped data centres, giving entities a standard, secure interface to open-source models, Mistral AI models and soon models trained on internal data; initial test phase entities included Hello bank! and the Group's general inspection.
Where does BNP Paribas use AI today?
| System | What it does | Status | Use case |
|---|---|---|---|
| LLM as a Service platform[1][2] | Shared, GPU-equipped infrastructure in the Group's data centres offering open-source and Mistral AI models through a standard interface; deployed first in test phases (Hello bank!, general inspection) and intended to roll out progressively across the Group. | Rolling out | Generative & agentic AI |
| LLM@CIB assistant[15][17] | The corporate and investment bank's generative AI assistant, available to more than 65,000 employees; Gemini models are to be integrated. | In production | Generative & agentic AI |
| 'My companion' Group AI platform for employees[12][7][9] | A Group AI platform that had onboarded 100,000 people by the Q2 2026 results; Bloomberg reported a progressive roll-out of an AI companion to all staff, and the Mistral announcement describes an employee assistant to summarise, reformulate and translate content, to be deployed in 2026. | Rolling out | Generative & agentic AI |
| HelloïZ virtual assistant at Hello bank![9] | The latest version, using generative AI (Mistral AI models) to understand and route requests with access to human advisers, has been deployed to more than one million clients since January 2026. | In production | Customer-facing chatbots |
| Nickel Assist[15] | Gemini is the foundation model for Nickel Assist, which helps 200 Nickel customer advisers find answers to customer requests from Nickel's procedures and operating guidelines. | In production | Generative & agentic AI |
| AI agents for corporate credit memos[15][16] | Planned deployment of agents on Gemini Enterprise for targeted CIB workflows, starting with the preparation of corporate credit memos and extending to sales, trading, research and structuring. | Announced | Credit scoring & underwriting |
| KYC and compliance with Mistral AI models[9][10] | On-premises generative AI for KYC steps where data sensitivity is high, automating low-value steps with expert validation of risks and decisions; Reuters reported Mistral is also used for compliance at BNP Paribas Fortis. | Rolling out | AML / KYC |
| ESG Assessment assistant[3] | An internal assistant built with ChatGPT on Microsoft Azure suggests answers and cites document sections for the ESG Assessment; used for around 35% of ESG Assessments, with 600 users. | In production | Generative & agentic AI |
| Inspection Générale audit assistants[6] | A generative-AI Virtual Assistant (2024) and, in 2025, IG Data Explorer, IG Datas and Analyzer and IG Scribe help auditors with analysis, translation, synthesis and audit-report formatting. | In production | Generative & agentic AI |
| Real-time transaction fraud detection[18] | BNP Paribas says AI analyses millions of transactions in real time to detect abnormal behaviour and prevent fraud, as part of more than 800 AI use cases. | In production | Fraud detection |
| Preparing for AI-driven vulnerability discovery[10] | The CIO said the bank is working to process and fix large volumes of vulnerabilities discovered at once; the Mistral partnership covers on-premises solutions for sensitive processes. | Announced | Cybersecurity |
| Generative AI for software development[1][12] | Generative AI lets developers create code and access development tools centrally; the Q2 2026 slides list industrialisation of the software development lifecycle including AI among five cost-saving levers. | In production | AI-generated code & coding agents |
| Group AI Committee, AI Risk Committee and responsible-AI principles[6][11][12][8] | The Group AI Committee approves the AI strategy and evaluates use cases against risk assessments; the AI Risk Committee monitors AI Act compliance, defines model attestation and analyses AI risk; five responsible-AI principles apply across the AI lifecycle; research runs through the FinAI-LAB joint laboratory with Télécom Paris and a collaboration with Mila. | In production | AI governance (general) |
Use-case links open the tracker's rules-by-use-case matrix: which documents govern each system type.
What has BNP Paribas disclosed in numbers?
| Metric | Value | As of |
|---|---|---|
| Total consolidated balance sheet (total assets, as reported) | €2,793.0bn[6] | Dec 31, 2025 |
| AI use cases already in place | More than 800[18] | Oct 5, 2026 |
| Employees with access to LLM@CIB | More than 65,000[15] | Sep 24, 2026 |
| People onboarded to the 'My companion' Group AI platform | 100,000[12] | Jul 23, 2026 |
| Employees who received AI-related training in 2025 | Nearly 80,000; more than 30,000 completed Introduction to GenAI[11] | Dec 31, 2025 |
| Annual cost savings (slides) | About €700 million average 2022–2026, rising to about €1 billion average 2027–2030[12] | Jul 23, 2026 |
| Share of next incremental efficiency gains expected from AI | Around half[5] | Feb 5, 2026 |
| Hello bank! clients with the HelloïZ virtual assistant | More than one million since January 2026[9] | May 26, 2026 |
| Nickel customer advisers using Nickel Assist | 200[15] | Sep 24, 2026 |
| ESG Assessments completed with the AI assistant | Almost 1,600 by 600 users; around 35% of all ESG Assessments[3] | Nov 12, 2025 |
“With LLM as a Service, we are creating a common technological foundation that allows entities to focus on business use cases, while relying on a shared and secure infrastructure. It is a common lever to innovate and improve the performance of certain applications.”[1]
Marc Camus · Group Chief Information Officer · Jun 5, 2025
“We’re investing where it actually makes sense, because it’s really expensive,”[5]
Jean-Laurent Bonnafé · Chief Executive Officer · Feb 5, 2026
“The game changer is the speed at which we have to address vulnerabilities and the scale.”[10]
Marc Camus · Group Chief Information Officer · May 26, 2026
“A multi-cloud and multi-model approach gives us the flexibility to select, for each case, the right model, infrastructure and level of control to meet our requirements for quality and security while delivering the best balance between cost and value.”[15]
Marc Camus · Group Chief Information Officer · Sep 24, 2026
“Our approach to AI starts with the business need.”[15]
Charles Holive · Chief AI Officer, BNP Paribas CIB · Sep 24, 2026
Who runs AI at BNP Paribas?
Charles Holive
Chief AI Officer, Corporate & Institutional Banking (named January 2026; formerly managing director at JPMorgan Chase and SVP of AI solutions and platforms at PepsiCo)[4][15]
Olivier Osty
Deputy Chief Operating Officer of BNP Paribas and Chief Executive Officer of Corporate & Institutional Banking[9]
Which regulators govern BNP Paribas's AI?
| Authority | Why it matters here | Documents |
|---|---|---|
| ECB | BNP Paribas is supervised within the Single Supervisory Mechanism; the ECB's 2026–28 priorities and its July 2026 letter on AI-enabled cyber threats bear on the Mistral cyber work and agentic roll-out. | SSM supervisory priorities 2026–28 ECB 'Dear CEO' letter on AI-enabled cybersecurity threats (SSM-2026-0301) |
| EBA | The EBA's mapping of AI adoption in EU banking is the benchmark for BNP Paribas's 800-plus use cases and its multi-provider model strategy. | EBA report: Rising application of AI in EU banking and payments (Sep 2025) |
| EU AI Act | BNP Paribas says its responsible-AI principles are implemented in compliance with the AI Act, and its AI Risk Committee monitors AI Act compliance; credit-memo, KYC and customer-assistant use cases sit near the Act's high-risk and transparency rules. | Regulation (EU) 2024/1689 EBA factsheet on the AI Act |
| Basel Committee | A stack built on Mistral AI, Google Cloud, open-source models and its own data centres is a third-party and data-aggregation question under the Basel principles. | BCBS Third-Party Risk Principles (Dec 2025) BCBS 239 |
| FSB | Agentic AI in a multi-cloud estate falls within the FSB's AI sound-practices consultation and third-party risk work. | FSB AI sound practices consultation (June 2026) |
| ACPR | French supervisor of BNP Paribas alongside the ECB; the 2025 Universal Registration Document records ACPR and SSM representatives attending the Board's meetings of 25 March and 16 December 2025. |
What does the public record suggest about BNP Paribas's AI strategy?
Analysis by BankingNewsAI from the sources cited on this page. Observations, not advice.
01
AI is now a cost-plan line, so the proof will be in the cost-to-income ratio
With savings targeted to rise from about €700 million to about €1 billion a year and AI named as one of five levers, investors will read the 2027 plan for how much of that is attributable to AI rather than shoring, organisation and procurement.
02
Sovereignty and scale are being balanced rather than chosen
A European partner (Mistral AI), a US hyperscaler (Google Cloud) and in-house infrastructure coexist, with sensitive data kept off the public cloud. The explicit multi-model stance suggests BNP Paribas intends to switch providers by use case.
03
Agent security is being specified before agents scale
The Google announcement stresses that each agent is authenticated, limited to the resources its task needs and monitored, which is the control vocabulary supervisors are converging on for agentic AI.
04
No headcount number is attached to AI yet
The bank quantifies savings (about €1 billion a year in 2027–2030) and the CEO describes AI as a way to cut running costs in support functions, but no AI-attributable headcount or FTE figure has been published.
What is BNP Paribas doing with AI?
BNP Paribas says it runs more than 800 AI use cases and has built an internal LLM as a Service platform (June 2025), renewed its Mistral AI partnership for three years (26 May 2026) and signed a five-year Google Cloud partnership for Gemini Enterprise and agentic AI (24 September 2026). Its Q2 2026 slides place AI at the heart of the 2027–2030 plan.
Does BNP Paribas partner with Mistral AI or Google?
Both. BNP Paribas has worked with Mistral AI since 2023, signed a first Group-wide contract in 2024 and renewed it for three years on 26 May 2026; on 24 September 2026 it announced a five-year partnership with Google Cloud to use Gemini Enterprise and Gemini models, starting with corporate and investment banking. It describes its approach as multi-cloud and multi-model.
Who leads AI at BNP Paribas?
Group CIO Marc Camus runs the technology platform, including LLM as a Service; Charles Holive is Chief AI Officer of the corporate and investment bank; Sophie Heller is Chief Transformation Officer of CPBS and an executive sponsor of the Mistral partnership. Strategy is set by a Group AI Committee, and the Q2 2026 slides say the Group AI strategic committees are chaired by the CEO, Jean-Laurent Bonnafé.
What AI tools do BNP Paribas employees use?
LLM@CIB is available to more than 65,000 corporate and investment bank employees, a Group AI platform called 'My companion' had 100,000 people onboarded by July 2026, and entity-level tools include an ESG Assessment assistant and Inspection Générale audit assistants. Nearly 80,000 employees received AI-related training in 2025.
Will AI cost jobs at BNP Paribas?
BNP Paribas has not published a headcount target for AI. CEO Jean-Laurent Bonnafé said on 5 February 2026 that AI would help deliver savings by streamlining procedures and cutting running costs across support functions, and the bank expects AI to deliver around half of the next incremental efficiency gains; its Q2 2026 slides raise targeted annual cost savings from about €700 million to about €1 billion for 2027–2030.
How does BNP Paribas govern AI?
Through a Group AI Committee that approves the AI strategy and evaluates use cases against risk assessments, an AI Risk Committee that monitors AI Act compliance and model attestation, five responsible-AI principles with a sixth on environmental footprint being added, and training at scale. Its US activities are not covered on this page.
- BNP Paribas provides its businesses with an LLM as a Service platform to accelerate the industrialization of generative AI use cases.
- BNP Paribas's gen AI startup incubator a step in AI journey
- BNP Paribas deploys AI to speed up ESG assessments
- Wells Fargo, BNP Paribas bolster AI leadership
- BNP Paribas vows more cost cuts as profit tops forecast
- 2025 Universal Registration Document and Annual Financial Report
- BNP Paribas Set to Roll Out AI Companion for All Staff
- AI research at BNP Paribas: innovating to reinvent finance
- BNP Paribas and Mistral AI extend their partnership to support the next phase of generative AI deployment within the Group
- BNP Paribas steps up Mistral partnership to bolster rapid AI defences
- Responsible AI at BNP Paribas: Innovating with Purpose and Confidence
- Second Quarter 2026 Results presentation (23 July 2026)
- “We are rolling out next-generation technology at scale to better serve our clients and further enhance our employees’ experience.”
- BNP Paribas leverages its leading franchises to advance a compelling growth trajectory through Europe's major transformations
- BNP Paribas and Google Cloud Announce New Partnership on Agentic AI and Cloud Innovation
- BNP Paribas CIB Strikes AI Deal with Google Cloud, Credit Memo/Res
- BNP Paribas forges agentic AI partnership with Google Cloud
- Tech & Artificial Intelligence FAQ (undated page, retrieved 2026-10-05)
When BNP Paribas 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 ET · free
plus every tracker, bank and agent page update, the morning after · leave any morning