Money 20/20 2025: Takeaways & News from Starling Bank and Google

In the fast-evolving world of fintech, clarity and confidence are emerging as the defining traits of true innovation.

At 2025’s Money20/20, conversations about AI in financial services, fraud detection, and cross-industry collaboration took center stage. For example, Starling Bank announced its partnership with Google Cloud, which lets it use Gemini’s AI models to fight digital fraud. Specifically, this is done through an in-app feature called Scam Intelligence. In addition, the initiative shows how banks and tech partners can work together to make fraud prevention more proactive and user-friendly. As a result, customers receive real-time help against scams, while Starling builds trust through clear and careful AI use. Moreover, the approach emphasizes ethical practices, trial-and-error testing, and close cooperation with regulators. Therefore, this collaboration sets a strong example of human-centered AI in banking. Finally, it demonstrates that innovation in fintech is not only about speed but also about trust and safety.

Unlike the overpromised declarations that often dominate fintech events, Starling’s approach stood out for its transparency and iterative design. Rather than unveiling a polished final product, Harriet Rees, Chief Information Officer at Starling, emphasized their test-and-learn model, a philosophy grounded in experimentation, explainability, and regulatory cooperation.

This adaptive framework not only reinforces customer trust but also highlights how responsible AI can serve as a differentiator in regulated industries.

What emerged from the discussion was a compelling case for human-centered AI deployment in banking. Rees illustrated how integrating advanced machine learning tools directly within customer touchpoints transforms risk mitigation into a shared, proactive experience.

As generative and agentic AI become embedded in financial operations, the leaders who prioritize ethics, usability, and partnership will drive meaningful progress.

Ultimately, the conversation underscored that innovation in fintech is no longer defined by speed, it’s measured by trust, transparency, and tangible outcomes. Through collaboration between technology giants and agile digital banks, the next phase of fintech evolution is focused on turning intelligence into intuition, where banking becomes safer, smarter, and nearly invisible.

Meet the Expert

Harriet Rees, Chief Information Officer at Starling Bank, is a seasoned leader in data science and AI strategy.

Since joining Starling over seven years ago, she has led initiatives that combine analytics, automation, and human insight to elevate customer experience and operational efficiency. Under her leadership, Starling has become known for its in-house innovation model, ethical data governance, and real-world AI applications, setting new standards for what modern banking can achieve. Her collaboration with Google Cloud on AI-powered fraud detection reflects her commitment to making technology both human-centered and future-ready.

The Big Idea

The central insight from Rees’ discussion is that AI in banking should enhance, not replace, human judgment. Starling Bank’s new Scam Intelligence feature exemplifies this balance, an AI tool that helps users identify fraudulent activity in real time, directly within their banking app.

By integrating Google’s Gemini model, the system analyzes customer-uploaded screenshots or listings for scam indicators, offering immediate guidance. This initiative reframes fraud prevention as a collaborative process between bank and customer, rather than a reactive service.

Rees highlighted that the key to innovation is iteration:

“These models can’t do all things. We need to make sure before we unleash them on customers, we’ve tried, tested, and understood their capabilities.”

This “experimentation-first” approach is redefining risk management and regulatory relationships. Instead of fearing compliance friction, Starling embraces open dialogue with regulators like the UK’s FCA and Bank of England’s AI Task Force, ensuring AI evolves responsibly within clear ethical boundaries.

Key Takeaways

Responsible AI is built on experimentation. Starling’s internal testing and human-in-the-loop oversight model enables safe scaling of machine learning.
Transparency builds trust. Open engagement with regulators allows innovation without compromising safety.
Data is a strategic asset, not a commodity. Starling’s philosophy centers on data usage, not data volume.
Collaboration drives capability. Google’s partnership provides cutting-edge tools, while Starling’s culture ensures practical application.
The next phase of fintech is invisible. As Rees notes, “The best technology in banking will feel effortless, and that’s the hardest thing to build.”

Tools, Strategies, or Frameworks Mentioned

Gemini AI Model (Google Cloud): Used for multimodal fraud detection and content synthesis within Starling’s app.
Scam Intelligence Feature: Allows users to upload images of suspicious listings for real-time scam analysis.
Internal AI Testing Environments: Starling uses in-house “sandbox” frameworks for experimentation before customer deployment.
Regulatory Collaboration: Engagement with the UK FCA and Bank of England AI Task Force ensures compliance and knowledge sharing across institutions.
Human-in-the-Loop Systems: A control layer that blends automation with human review, reinforcing accuracy and accountability.

Final Thoughts

Harriet Rees’ vision captures the essence of where fintech is headed, toward quiet innovation that prioritizes customer safety and strategic integrity over spectacle. By anchoring AI strategy in clarity, collaboration, and control, Starling Bank demonstrates how modern banking can remain both technologically advanced and deeply human.

“The future isn’t fintech vs. banks, it’s fintech with banks.”

This collaborative mindset is shaping a financial ecosystem where technology disappears into the background, empowering customers, protecting assets, and redefining what trust looks like in the age of AI.