Banks eye responsible AI with global spend set for USD $67bn
AI has moved from pilot projects to a board-level mandate. A new report from SAS, featuring leaders at Banorte, Intesa Sanpaolo, Millennium BCP, and Old National Bank, shows how banks are building responsible, profitable AI programs. With IDC projecting banking AI spend to near USD $67 billion by 2028, the message is clear: treat AI as strategy, not a side quest.
1) Business alignment
Bank leaders are explicit: AI must sit inside enterprise strategy and P&L conversations. "Innovation and AI must be recognised as a pillar of the institution's strategy," said Abraham Izquierdo, Managing Director of Trading and Treasury Risks at Banorte. "Leadership from the top is nonnegotiable."
The risk office is already putting this into practice. "We are developing models to help us better prepare for adverse scenarios in the next five years. If an economic crisis arises, we can more effectively mitigate the effects of a rise in defaults," said José Miguel Pessanha, Chief Risk Officer at Millennium BCP.
2) People-focused approach
Technology scales decisions, but people set direction. "AI tools cannot definitively answer key questions: How will I minimise the impacts for the bank? What will be the risk strategy afterwards? This comes down to the creative problem-solving abilities of leadership," said Pessanha.
Culture is compounding value. "By recognising data-driven successes, we create a culture where data is valued as a strategic asset," said Andrea Cosentini, Head of Data Science and Responsible AI at Intesa Sanpaolo. The bank is also using historical loan data to open credit for underserved segments, promoting financial inclusion.
3) Core infrastructure
Advanced use cases need clean, connected data and compliant pipelines. "We saw the need to harmonise our data to service different needs - from provisioning to capital calculations, liquidity calculations and interest rate risk," said Pessanha. "Walking before you run is essential."
Engineering readiness is speeding delivery. "GenAI is exceptionally helpful in developing a full code base. With this insight, we're driving a new wave of innovation," said Andrew McCammack, Data Science Officer at Old National Bank. Izquierdo added: "AI and cloud deployments must be multi-year and step-by-step. Recognise your long-term ambition, but set short-term, precise goals. Commit to a pace that maintains compliance and builds trust."
4) Empowering innovation
Giving teams access to AI tools is reducing manual work and creating time for analysis. Old National automated loan data processes: "Microsoft Excel sheets don't even exist anymore," said McCammack. "Now, people have time for deep analysis that's more rewarding and valuable to the business."
Curiosity is the engine. "We encourage an environment where asking 'why' is valued. Data scientists can then design experiments and models to test hypotheses and validate - or disprove - assumptions, potentially revealing breakthrough solutions," said Pessanha. Izquierdo added: "We can make strategic decisions for the short term, medium term and long term based on facts, data and modelling."
5) Continuous learning
The leaders who win keep listening and iterating. "We're looking to connect more with academics and startups to understand what they are focusing on," said Pessanha.
Development models are shifting. "We've started pursuing a strategy where we are building a whole suite of light applications, where AI is used to build the source code [...]. If you want to scale that and roll it out to the enterprise, you need a full code base, and GenAI is exceptionally helpful in developing it," said McCammack. Cosentini added: "GenAI can fuel innovation by generating new ideas and solutions... its full potential is realised only when guided by human oversight, ensuring that creativity, responsibility and strategic judgment remain at the centre of the process."
Izquierdo also pointed to security: "Generative AI will play an important role for us in strengthening our cybersecurity posture and fostering business continuity."
The foundation: trust and governance
"We've all heard the phrase 'trust is earned in drops and lost in buckets.' For banks, that saying rings especially true," said Stu Bradley, Senior Vice President of Fraud, Risk and Compliance Solutions at SAS. "AI can help banks strengthen trust when deployed responsibly, but without strong governance and guardrails, the risks rise faster than the rewards."
Executive actions for the next 90 days
- Name an accountable owner for AI who reports into both business and risk, with clear ROI and risk metrics.
- Prioritise three use cases with measurable value and compliance requirements; kill or pause low-value experiments.
- Stand up a data harmonisation plan across finance, risk, treasury, and customer analytics with shared definitions.
- Codify GenAI policy: human-in-the-loop review, data privacy, secure coding standards, and vendor controls.
- Expand model risk management to cover ML and GenAI: documentation, testing, monitoring, and bias reviews.
- Invest in skills: train leaders, risk teams, engineers, and product owners on AI literacy and decision-making.
- Build a lightweight delivery pipeline for AI apps: versioning, CI/CD, observability, and rollback plans.
- Create a trust dashboard for the board: value delivered, model performance, incidents, and customer outcomes.
If you are building team capability and need structured upskilling by role, explore our curated programs at Complete AI Training.
Your membership also unlocks:
