Torna alle notizie
ImpresaAI Understanding briefing

BNP Paribas partners with Google Cloud for AI while keeping sensitive data on-premise

BNP Paribas has entered a five-year agreement to integrate Google's Gemini AI models into its operations, while explicitly excluding sensitive customer data from the public cloud.

4 min readRead the linked source
Source-provided image accompanying BNP Paribas partners with Google Cloud for AI while keeping sensitive data on-premise
Riferimento alla fonteFonte registrata
Editore
finance.biggo.com
Collegamento alla fonte
finance.biggo.comhttps://finance.biggo.com/news/1b25ac28-b85c-4acd-9901-75986afe0849
Tipo di fonte
Fonte collegata: lo stato di fonte primaria non è stato stabilito.
ContestoComprendilo in 60 secondi

Inizia qui

Termini chiave

IA generativa
Sistemi di intelligenza artificiale che producono nuovi contenuti come testo, immagini, audio, video o codice.
Robustezza
La capacità di un modello di mantenere le prestazioni in condizioni di rumore, cambiamenti o input contrastanti.
Mettiti alla provaQuiz sugli agenti IA

Cosa è successo

BNP Paribas has announced a five-year partnership with Google Cloud to deploy AI agents across its global operations. The bank plans to utilize Google's Gemini Enterprise models to automate internal tasks, including drafting credit memos and supporting sales and trading teams. Despite the collaboration, the bank confirmed it will maintain its existing on-premise infrastructure for highly sensitive data, such as medical information from its insurance division. The bank will continue to employ a multi-cloud and multi-model strategy, maintaining existing relationships with other providers like Mistral.

BNP Paribas, one of Europe's largest lenders, announced a five-year strategic partnership with Google Cloud on Wednesday. The agreement provides the bank access to Google's public cloud infrastructure and the Gemini family of AI models.

The bank intends to develop 'agentic AI' to assist with internal workflows, such as research, structuring, and drafting corporate credit memos. A pilot project is already active within the bank's Nickel payment business, where 'Nickel Assist' helps customer advisers navigate internal procedures.

Group CIO Marc Camus emphasized that the deal does not signal a move away from on-premise systems. The bank will continue to use a multi-cloud and multi-model strategy, selecting different providers based on specific security, cost, and performance requirements.

Google Cloud CEO Thomas Kurian stated that the partnership focuses on deploying intelligent agents to streamline operations and risk assessment. Google confirmed that BNP Paribas retains full ownership of its data, which will not be used to train or refine Google's underlying models.

Dettagli della fonte: finance.biggo.com ↗

Perché è importante

This partnership highlights the tension between the rapid adoption of in finance and the stringent data sovereignty requirements of European banking regulators. By explicitly walling off sensitive data from the public cloud, BNP Paribas is establishing a hybrid operational model that allows for AI innovation without compromising risk management. This approach serves as a blueprint for other large financial institutions navigating the balance between leveraging third-party AI capabilities and maintaining strict control over regulated, confidential information.

The financial sector faces significant pressure to adopt to remain competitive, yet it is constrained by strict data privacy and sovereignty laws. BNP Paribas's strategy of keeping sensitive data on-premise while using public cloud for less critical tasks addresses these regulatory hurdles directly.

By maintaining a multi-model approach, the bank avoids dependency on a single vendor, allowing it to pivot between providers like Google and Mistral as AI capabilities evolve. This flexibility is critical for a global institution operating in 64 countries with varying regulatory landscapes.

The bank's commitment to keeping sensitive data off the public cloud reflects a broader industry trend where banks act as 'sovereign' users of AI, prioritizing internal control and auditability over the convenience of full-scale cloud migration.

Interactive Mechanism

Meccanismo interattivo: come funziona realmente

Esplora la tecnologia alla base di questo sviluppo in modo interattivo.

Agent Lifecycle Stage:
1
User Intent & Planning: "Audit customer refund request #4092 and settle payment."
2
Tool Calling: Emits structured JSON call crm_get_transaction(id='4092').
3
Guardrail & Verification:🛡️ Paused: High-value action requires human operator sign-off.
4
Final Settlement: Refund recorded, email receipt dispatched, and audit log stored.
Core takeaway: An AI agent is not just a language model—it is a closed loop of planning, tool invocation, and environment feedback. Production systems require self-healing retries and strict human approval guardrails.
Verifica concettuale interattiva+10 Points
AI Agents Quiz

What most distinguishes an AI agent from a basic chatbot?

Cosa guardare dopo

Observers should monitor how BNP Paribas manages the integration of these AI agents across its 64-country footprint and whether the bank's 'multi-model' approach effectively mitigates vendor lock-in. Additionally, the success of the initial rollout in the Corporate & Institutional Banking division will be a key indicator of whether these agents can handle complex, high-stakes financial workflows while meeting internal security standards. Future updates may reveal how the bank navigates evolving European regulatory expectations regarding third-party AI risk.

The effectiveness of the bank's 'authenticated' AI agents will be a primary metric for success. The bank stated these agents will only have access to resources necessary for their specific tasks, which will be subject to ongoing monitoring.

The bank's ability to scale these tools from the Nickel payment unit to more complex institutional banking divisions will test the of its security framework.

Future developments will likely focus on how the bank balances the cost-efficiency of Google's cloud infrastructure against the maintenance costs of its retained on-premise systems.

Guide e quiz correlati

Agenti dell'intelligenza artificialeSpiegazione dei modelli di intelligenza artificialeEtica dell'IAMetti alla prova ciò che sai: prova un quiz gratuito sull'intelligenza artificialeCerca un termine AI nel nostro glossario
Lo hai trovato utile?