Tech and the City – AI in Financial Services – techUK and TheCityUK Half-Day Summit

Opening Keynote – Edmund (Ed) Towers, Head of Advanced Analytics & Data Science Units, FCA

Ed set the scene by reaffirming the FCA’s longstanding role as a global innovation leader, from launching the world’s first regulatory sandbox in 2016 to today’s AI-focused support services:

1. Supercharged Sandbox – enabling early-stage experimentation and proof-of-concept development
2. AI Live Testing – allowing firms to test AI solutions with real users in a live market setting

Both sit within the FCA’s dedicated AI Lab, which has already seen strong participation, with 50+ firms engaging in its first cohort. The dominant theme so far: a surge in generative AI experimentation, often combined with traditional machine learning and emerging agentic systems.

---

On regulation, Ed reaffirmed that the UK will maintain its principles-based, outcomes-focused approach. Core frameworks such as the Consumer Duty, Senior Managers and Certification Regime, and operational resilience and cyber rules continue to apply to AI systems. His message was clear:
The UK has the ingredients to lead globally in AI-enabled financial services — but continuing to build confidence in safe, reliable deployment is the key next step.

---

Second Keynote – Alexon Bell, Co-Founder & Chief Product Officer, Quantexa

Alexon highlighted the transformational moment facing the financial technology ecosystem and the importance of collaboration between regulators, FS firms and technology providers.

He stressed the need for realism in AI strategies: despite the focus on frontier AI, around 80% of AI in financial services is still traditional machine learning, and only 4% of generative AI Proof-of-concepts (PoCs) make it to production. Many AI initiatives fail simply because the wrong type of AI is used for the wrong problem — or because explainability and end-user needs aren’t designed in from the start.

A central pillar of Alexon’s keynote was data. AI cannot succeed without clean, organised, high-quality data. As he noted, data must be refined like oil before it becomes useful. The famous self-driving car example — where models trained in Europe failed to recognise kangaroo movements in Australia — underscored the importance of diverse training data and context-specific design.

The takeaway: strategic focus, data discipline and a willingness to fail forward are key to scaling AI in financial services. Successful AI implementations can be replicated across the ecosystem, helping firms accelerate adoption and build market share.

---

Panel Discussion – Challenges & Opportunities for AI in Financial Services

Panellists:

Following the keynotes, industry leaders joined the stage for a discussion on the challenges and opportunities for AI in financial services. The panellists were:

• Sushil Saluja, Technology & Innovation Lead, City of London Corporation (Panel Chair)
• Jeremy Donaldson, Managing Director – Head of Financial Services, DXC Technology (techUK Financial Services Council Chair)
• Suchitra Nair, Head of EMEA Regulatory Strategy, Deloitte (TheCityUK Technology and Innovation Group Chair)
• Chris Knox, Global Financial Services Regulatory Director, Microsoft
• Sarah Self, AI Director, Aviva

The panel explored the major trends shaping AI adoption across the sector. Key focuses of the discussion were:

1. AI is being deployed across organisations — but nervousness remains

Businesses increasingly see AI as a way to enhance almost every process. Yet concerns persist about reputational, operational and financial risks if systems fail.

2. Data remains the biggest barrier to scale

Echoing Alexon’s keynote, panellists agreed that messy, siloed data is slowing AI roll-out and preventing scaling. Bright spots include strong adoption in payments and fast-emerging opportunities around code modernisation for legacy systems.

3. Global adoption rates vary sharply

• US: leading due to capital availability, competitive pressure and high pilot tolerance
• China: releasing new models at pace
• Europe/Singapore: more measured, with focus on guardrails and governance

4. Regulation: UK approach broadly right, with some gaps

Panellists supported the UK’s principles-based regime but flagged the need for:

• Clearer guidance on agentic AI within model risk management
• Addressing validation challenges posed by large models, where FS firms lack visibility into training data or architecture

The consensus: the biggest blockers to AI adoption aren’t regulatory — they’re organisational. Data quality, siloed operating models and inconsistent governance frameworks are the main friction points.

5. Operationalising AI: Start small, stay focused

Many AI programmes have failed because they spanned too many teams and lacked coherent governance. Firms now recognise the need to:

• Pick a small number of high-value use cases
• Set clear business objectives
• Build unified frameworks and data foundations

---

Closing Remarks – James Challinor, techUK

James closed the summit by reaffirming the significant opportunity for technology firms to support the financial services sector’s growth mission. While this event focused on AI, techUK will also explore other transformative technologies — including distributed ledger technology and quantum computing — through its Financial Services Programme in 2026.

For more information or to get involved in techUK’s work across financial services and advanced technologies, please contact [email protected].

---

---