AI becomes banking's connective tissue as firms boost security, scale cloud, and see measurable gains

AI is now finance's connective tissue

AI is no longer a debate in banking. It's infrastructure. Finastra's latest State of the Nation report shows all but 1% of UK financial services firms use AI, and only 2% of global respondents say they don't use it at all.

The report calls AI the sector's "connective tissue" - the intelligence layer linking data, channels and services into a coherent, responsive system. In short: AI is how modern finance operates.

Where AI delivers value today

• Risk management and fraud detection: 71% currently running or piloting.
• Data Analysis and reporting: 71% using AI to accelerate insight and close reporting cycles.
• Customer service and support assistants: 69% improving resolution times and containment rates.
• Document intelligence and management: 69% extracting, classifying and validating at scale.

Finastra's view is blunt: AI now sits at the heart of financial innovation, from real-time fraud blocks to intelligent underwriting and dynamic customer engagement. The conversation has moved from "should we" to "where does it create measurable value and how do we run it responsibly?"

Execution priorities for the next 12 months

• AI-driven personalisation: Offers, pricing and servicing that adapt in real time.
• Agentic AI for workflow automation: Systems that take actions, not just make predictions.
• Model governance and explainability: Controls, monitoring and auditability that stand up to regulators.

Over 1,500 senior executives across 11 regions contributed to the survey, pointing to a sector that has moved beyond pilots and is now executing.

Security spend jumps 40% - prepare your stack

AI is also pushing up security budgets. Firms expect an average 40% increase in 2026 security spend, driven by escalating digital threats linked to AI. The biggest pain points: constantly evolving risks (43%) and the complexity of AI deployment itself (40%).

• Refresh your threat model for AI-era attacks: prompt injection, model theft, data poisoning, synthetic fraud.
• Isolate high-risk models and data with strict identity, secrets, and key management.
• Stand up red-teaming and continuous testing for both traditional and AI-specific vectors.
• Instrument end-to-end observability: model drift, anomaly rates, abuse signals and human override rates.

Modernise for AI at scale

Nine in ten firms plan to modernise this year, in large part to scale AI. Cloud is central, with 29% prioritising adoption and 84% already using some cloud solutions. As the report puts it, "Cloud is no longer a destination; it is the operating environment for modern finance."

• Standardise on a multi-cloud data platform with governed access, lineage and feature stores.
• Adopt MLOps: versioning, CI/CD for models, automated testing, and rollback plans.
• Shift to event-driven, API-first architectures so models can act in real time.
• Rationalise vendors and reduce hidden latency, egress and integration costs.

ROI is improving - and fast

Lloyds Banking Group reports a clear step-up: 59% of firms saw Productivity gains in 2025 (vs 32% in 2024). 21% say AI is directly driving business growth (vs 8% the prior year). 33% report better customer experiences, and the same share now has deeper customer insights (both up from the teens a year earlier).

• Track ROI with precision: productivity deltas, fraud loss avoided, NIM uplift, call deflection, STP rate, time-to-yes.
• Target compounding wins: use shared services (identity, data quality, model monitoring) across multiple use cases.

A practical playbook for finance leaders

• Pick 3-5 needle-movers: e.g., real-time fraud prevention, credit decisioning, KYC automation, collections optimisation.
• Stand up governance now: risk taxonomy, model inventory, bias testing, explainability standards, approvals workflow.
• Upgrade security: least privilege by default, secret rotation, model isolation, human-in-the-loop for high-risk actions.
• Modernise data: golden sources, reference data cleanup, PII minimisation, retention rules, and lineage you can audit.
• Industrialise MLOps: reproducibility, monitoring SLAs, drift alarms, champion-challenger, rollback in minutes.
• Close the skills gap: upskill risk, compliance and ops on AI basics; hire AI product owners; train frontline teams.
• Tighten vendor due diligence: model cards, security posture, data residency, indemnities and exit strategy.
• Regulatory alignment: map controls to frameworks like the NIST AI RMF and the EU AI Act.
• Change management: update policies, incentives and KPIs so teams adopt, not bypass, AI-infused workflows.

Execution over experimentation

Technology choices now sit at the centre of trust, resilience and customer experience. As Finastra's leadership notes, the sector has moved past trials and into delivery - with higher standards for safety, reliability and personal relevance every time.

If your roadmap doesn't reflect this shift, you'll feel it in fraud losses, opex, and customer churn before year-end. If it does, 2026 becomes the year AI connects your data, channels and services into one responsive system - and your numbers will show it.

Useful resource

• Curated AI tools for finance - practical options to accelerate risk, fraud, reporting and service use cases.