FSB Hosts Global Event on Best Practices for AI in Financial Institutions

The Financial Stability Board (FSB) has taken another important step in building a global framework for the responsible adoption of artificial intelligence in the financial sector. The organization held an international virtual event aimed at discussing healthy practices that financial institutions should follow when incorporating AI technologies into their operations.

The initiative reinforces a growing trend among regulators and international bodies: artificial intelligence cannot be adopted indiscriminately by the global financial system without clear guidelines, risk control mechanisms, and a robust governance structure.

---

What is the FSB and Why Does it Matter?

The Financial Stability Board is an international organization created after the 2008 financial crisis, composed of representatives from central banks, finance ministries, and regulatory authorities from the world’s largest economies, including G20 members. Its main mission is to monitor and recommend actions regarding the global financial system, aiming to prevent systemic crises.

When the FSB speaks about artificial intelligence, the market pays attention. The organization has the power to influence regulatory policies in dozens of countries, making this event a significant milestone for banks, fintechs, insurers, and asset managers worldwide — including Brazil.

---

Why Does AI in the Financial Sector Need Regulation?

The adoption of artificial intelligence by financial institutions has grown exponentially in recent years. Banks utilize AI for:

• Credit analysis and risk scoring
• Real-time fraud detection
• Customer service via chatbots and virtual assistants
• Automated investment portfolio management (robo-advisors)
• Compliance and regulatory monitoring (RegTech)
• Pricing of financial products and insurance

However, this accelerated expansion brings significant risks that cannot be overlooked. Among the main challenges identified by global regulators are:

1. Algorithmic Bias Risk

AI models trained on historical data can perpetuate or amplify existing discriminations in society. A credit granting algorithm, for example, may unintentionally disadvantage specific population groups based on historical exclusion patterns.

2. Model Opacity ("Black Box")

Many deep learning models are extremely difficult to interpret, even by their creators. This creates a serious explainability problem: how can a bank justify to a customer the denial of a loan if even its own technicians cannot fully explain the model’s reasoning?

3. Systemic Risk and Correlation

When multiple financial institutions use similar AI models, there is a risk that they will all make similar decisions at the same time, potentially amplifying market movements and creating systemic instability — exactly the kind of scenario the FSB was created to prevent.

4. Cybersecurity and Adversarial Attacks

AI systems can be vulnerable to specific attacks designed to deceive them, known as adversarial attacks, posing a threat to the operational integrity of institutions.

5. Third-Party Dependence

A large portion of the AI solutions used by banks comes from external vendors — often just a few large technology companies. This concentration of dependency can create single points of failure in the global financial system.

---

What the FSB Proposes: Recommended Practices

The FSB's event focused on the dissemination and discussion of sound practices that financial institutions should adopt. Based on the documents and guidelines the organization has been developing, the key pillars include:

Governance and Accountability

The FSB recommends that institutions establish clear governance structures for the use of AI, including the definition of human responsibilities for decisions made by automated systems. The concept of "human in the loop" is central to this approach.

Transparency and Explainability

Institutions must be able to explain the decisions made by AI algorithms, both to regulators and to the affected customers. This requires the use of explainable AI (XAI) techniques and rigorous documentation of the models used.

Continuous Risk Management

AI should not be treated as a "set it and forget it" product. The FSB emphasizes the need for continuous monitoring of models, including regular performance testing, checking for data drift, and periodic audits.

Security and Resilience

AI systems should be developed with security by design, incorporating protections against cyber attacks and ensuring that failures do not compromise the operational stability of institutions.

Third-Party Management

Given the high level of dependence on external technology vendors, institutions should implement rigorous due diligence on the AI solutions they acquire, ensuring that these vendors also follow responsible practices.

---

Impact for Brazil and the Brazilian Financial System

Brazil has one of the most technologically advanced banking sectors in the world. The Central Bank of Brazil has already shown interest in the regulation of AI, and initiatives like Open Finance and Pix have already incorporated elements of intelligent automation into their infrastructure.

FSB guidelines are likely to directly influence the Brazilian regulatory framework. Banks like Itaú, Bradesco, Nubank, and XP — which already utilize AI extensively — will need to pay attention to the organization's recommendations to maintain compliance with international standards, especially in their cross-border operations.

Additionally, Brazil's fintech ecosystem, one of the most vibrant in the world, will also be impacted. Financial startups that rely on AI as a competitive differential will need to balance innovation with the increasing regulatory requirements.

---

A Turning Point for Financial AI

The FSB event represents much more than a technical meeting among regulators. It signals that the world has reached a turning point in the relationship between artificial intelligence and the global financial system.

The central message is clear: AI is welcome in the financial sector, but it must be adopted responsibly, transparently, and under appropriate oversight. Enthusiasm for the possibilities of technology cannot overshadow the need to protect consumers, ensure system stability, and maintain public trust in financial institutions.

For industry professionals, the recommendation is unequivocal: investing in AI governance is not just an emerging regulatory obligation — it is a strategic competitive advantage in a world where trust is the most valuable asset.

---

The Financial Stability Board regularly publishes reports and public consultations on topics related to global financial stability. Documents on responsible AI practices are available on the organization's official website at fsb.org.