On June 25, 2025, in Warsaw, together with the Polish Bank Association, we had the pleasure of co-hosting another edition of the Technology Breakfast organized by the Banking Technology Forum. This time, the focus of the meeting was AI governance in the context of upcoming regulations such as the EU AI Act, as well as practical approaches to managing AI-based systems in the financial sector.
The Growing Importance of AI Governance in Banking
Managing artificial intelligence in a safe, transparent, and compliant way is becoming a necessity for banks and financial institutions. As the adoption of AI solutions—such as predictive models and large language models (LLMs)—accelerates, several key questions emerge:
How can compliance with the AI Act be ensured?
How can risks related to automated decision-making be controlled?
How can AI be implemented responsibly without stifling innovation?
These were the exact issues we addressed during the event—both in the expert presentations and the live demonstration sessions.
A Practical Approach to AI Act Compliance
One of the central topics of the meeting was the AI Act—the upcoming EU regulation on artificial intelligence. As our CEO, Krzysztof Goworek, pointed out, AI—much like space technologies—requires advanced oversight, and in banking, there is no room for error. Transparency, safety, and legal compliance are absolute priorities.
Krzysztof also emphasized that the AI Act can become a catalyst for safe innovation, rather than merely a constraint. Participants agreed that a unified regulatory framework at the EU level helps level the competitive playing field, although concerns were raised about so-called “gold-plating”—the risk of national regulators overinterpreting the rules.
He highlighted that the AI Act does not necessarily mean restrictions—it can actually help streamline processes and build trust in AI technologies among both customers and regulators.
I would argue that the fact we have the AI Act in Europe—and approach artificial intelligence differently from the U.S. or China—does not have to be a drawback. On the contrary, it might be our competitive advantage, driving safe and sustainable innovation. Krzysztof Goworek CEO of TUATARA
The discussion also addressed real-world challenges such as Shadow AI—cases where employees use AI tools (like ChatGPT) without the organization’s knowledge, potentially risking data leaks. Equally concerning are instances of embedded AI, where AI systems are integrated into technical infrastructure in a way that may escape oversight or regulatory auditing.
watsonx.governance – A Tool for Effective AI Lifecycle Management
In the second part of the event, Michał Trzęsiok, IBM Technical Specialist, presented the IBM watsonx.governance platform—a solution designed to support comprehensive AI system management within organizations. The platform covers the full AI lifecycle: from defining use cases, through risk assessment and regulatory compliance, to model monitoring and audit readiness.
IBM watsonx.governance was built with highly regulated industries in mind—such as banking and insurance—where organizations must demonstrate full control over the AI solutions they implement.
Key features and capabilities of the platform:
Creating and managing AI use cases: Each AI project is treated as a separate use case, with assigned roles, responsibilities, and full documentation at every stage.
Assessment of legal, ethical, and operational risks: With the help of the Risk Atlas tool, users can identify potential threats associated with a given solution – from bias risks and data drift to social and reputational risks.
Compliance with regulations (AI Act, KNF, internal policies): Built-in forms and decision-making paths make it possible to verify whether a project complies with legal requirements and automatically classify cases as high-risk in accordance with the AI Act.
Model performance monitoring and system oversight: The platform tracks quality metrics, monitors changes in input data (data drift), detects potential generative errors (e.g. hallucinations in LLMs), and evaluates compliance with fairness criteria (bias/fairness).
Automatic documentation and audit trail (AI Fact Sheet): Each AI project is supported with a set of documents including decision history, training data, test results, and a list of responsible individuals. This is a critical aid in both internal and external audits.
Business-friendly interface: Thanks to the dedicated management dashboard, business users can oversee project progress without engaging in technical details – the entire process is available without the need for coding.
The presentation demonstrated how IBM’s platform can help financial institutions not only meet regulatory requirements, but also effectively manage risk and build trust in AI systems – both internally and externally.
How to Responsibly Shape the Future of AI in Finance?
The Technology Breakfast dedicated to AI Governance demonstrated that the banking sector in Poland not only understands the significance of upcoming regulations, but is also actively seeking ways to turn them into a real competitive advantage. Among the participants, there was a shared belief that the upcoming changes – although challenging – can strengthen trust in AI-powered solutions, both within organizations and in the eyes of customers and regulators.
One key takeaway was that responsible AI implementation is not a one-off initiative, but a continuous process – one that requires clearly defined roles, ongoing monitoring, and tools that support compliance with an evolving regulatory landscape. Solutions such as watsonx.governance allow financial institutions not only to meet the requirements of the AI Act, but also to effectively manage risk and increase the transparency of their AI systems.
The discussions held during the event – from risk classification and shadow AI to ethical concerns – confirmed that today’s banking sector needs more than just technology. It needs the right competencies, processes, and a culture prepared for responsible automation.
One thing is certain: as AI plays an increasingly important role in finance, the true value it brings to business and customers will depend on how well we govern it.
Looking to streamline AI governance in your organization?