Banks are giving AI agents more work while keeping a close eye on how far they can go

What's happening

Banks and financial firms are moving from AI that only suggests actions to agentic AI that can execute tasks: issuing refunds, flagging transactions, drafting customer responses, triggering KYC steps, calling internal APIs and even moving money. A recent survey shows 52% of financial-services respondents are actively adopting agentic AI, while European regulators note more than 85% of supervised banks use AI already. Institutions say agents can produce more consistent, exhaustively documented investigations for financial crime, but teams struggle to prove to compliance exactly what an agent did, when, and which policy or model produced that action. At the same time, legacy systems and fragmented data are limiting how much AI surveillance and automation actually change operations, so banks are adopting agents carefully and keeping tight governance around execution boundaries.

Why it's trending

Because agentic AI is now capable of execution (refunds, KYC triggers, moving money), banks must balance operational gains with compliance, auditability and legacy-system constraints, making governance the hot issue today.

SignalHolding at its usual pace, confirmed across 2 independent source types.