Most Banks Use AI Agents, but Few Have Controls to Scale Them

Most major banks now permit AI agents in production for selected use cases, but only a small minority have comprehensive systems to monitor and control them, according to a Boston Consulting Group benchmark. The report finds that governance maturity is linked to faster, broader AI adoption rather than slower deployment.
- 73% of banks permit AI agents in production for selected use cases.
- At 88% of banks, agents make up 10% or less of the AI portfolio.
- Only 8% have comprehensive monitoring and control frameworks covering agents.
- 46% reported an AI-related issue or incident in the past 12 months.
- Banks with more mature risk management are 2.5 times more likely to be advanced AI adopters.
AI agents remain limited in banking
BCG's AI Safety and Governance in Banking 2026: Why Operationalizing Governance Gives Banks an Advantage in Scaling AI benchmark examined 26 global and domestic systemically important banks. It found that 92% are piloting or permitting agentic AI, while 73% permit agents in production for select use cases.
Despite that progress, agents remain a small part of most banks' AI activity. At 88% of the banks surveyed, agents account for 10% or less of the AI portfolio. None permit broad production use, suggesting that banks are limiting deployment while they develop safeguards for systems that can plan and execute multistep tasks, take actions, and adjust their approach without human input at every step.
Only 8% of banks have a comprehensive, standardized monitoring and control framework that explicitly covers agents. Meanwhile, 54% have defined restrictions or prohibitions on agentic AI in their risk-appetite frameworks. BCG says more than half are relying on such restrictions while broader monitoring and control capabilities are being built.
AI incidents highlight the operational gap
The benchmark indicates that AI-related risk is already affecting banks. Nearly half, or 46%, reported an AI-related issue or incident in the past 12 months. More than a quarter also said their AI systems were falling behind their promised business case, showing that the consequences can include lost value as well as major risk events.
Basic governance structures are more widely established. According to BCG, 96% of banks manage AI risk through existing risk frameworks, 73% have defined their appetite for AI risk, up from 57% in 2025, and 77% have a dedicated AI policy.
However, day-to-day implementation remains incomplete. Fewer than half have a centralized, tool-based AI inventory, and none have a comprehensive human-oversight framework consistently implemented across AI systems. Anne Kleppe, BCG's global lead for AI safety and governance and a report coauthor, said the challenge is turning governance from a framework into an operating capability.
More mature governance is linked to faster scaling
BCG's findings challenge the assumption that stronger governance necessarily slows AI adoption. Banks with more mature risk management are 2.5 times more likely to be advanced AI adopters.
Banks that use standardized, previously approved patterns for lower-risk AI applications are nearly twice as likely to move those applications from idea to production in under six months. BCG says standardizing repeatable use cases can allow banks to reserve deeper reviews for novel and higher-risk applications.
For banks, the immediate implication is that scaling agentic AI will depend on extending existing governance foundations into consistent technical and operational controls. Matteo Coppola, global leader of BCG's Risk and Compliance practice and a report coauthor, said the benchmark shows that governance and speed need not be a tradeoff. The report positions the current limited use of agents as an opportunity for banks to strengthen monitoring before agents become more deeply embedded.
PR Newswire: Operationalizing AI Governance Gives Banks an Advantage in Scaling AI


