BankingNewsAI Daily Brief · Saturday, March 28, 2026
Banking AI
Financial institutions & fintech technology
Bank of England updated outsourcing and third‑party risk expectations for payment systems and CCPs
The Bank of England issued updated supervisory statements on outsourcing and third‑party risk management for recognised payment system operators/specified service providers and for central counterparties. This matters because AI adoption is pushing more critical workloads into external cloud/model vendors, and the BoE is sharpening expectations around governance, resilience, and oversight for those dependencies.
Action
Tighten third‑party controls now by mapping AI/model vendors into your critical service taxonomy, stress-testing concentration and exit plans, and ensuring contract language covers audit rights, operational resilience metrics, and incident reporting aligned to BoE expectations.
FCA is accelerating its own AI use as consumers increasingly use AI to pick financial products
The UK FCA’s 2026/27 work programme says it will ramp up internal AI to strengthen supervision, explicitly tied to evidence that consumers are relying on AI tools when choosing financial products. The regulator is signaling it will modernize how it detects advice/suitability risks and may scrutinize AI-influenced customer journeys more aggressively.
Action
Assume supervision will get more data-driven: instrument and preserve explainable decision trails across digital sales/advice funnels, and proactively test how third-party consumer AI tools might mis-state your product terms or suitability prompts.
General AI
Large language models & AI infrastructure
SAP agreed to acquire Reltio to make enterprise master data “AI-ready” across SAP and non-SAP systems
SAP announced an agreement to acquire Reltio, an MDM provider, positioning it as a way to unify and govern customer/product data for AI use across heterogeneous estates. The key change is that major ERP vendors are treating data mastery as a first-class AI prerequisite, not back-office plumbing.
Action
Accelerate your own “AI-ready data” program: prioritize entity resolution, lineage, and permissions at the master-data layer (customer, counterparty, product) because model quality and auditability will bottleneck on data governance, not prompts.