Skip to content
AI ConnectPowered by VELENTIS
AI-generated2 min

SAS and Microsoft Study Finds Banks Lack Controls for Autonomous Risk AI

A global survey of 300 banking executives reveals major governance gaps: while firms expect autonomous credit decisions by 2031, many lack kill switches or continuous real-time monitoring.

This article was AI-generated and published automatically. Context, labelling and all sources at the end of the article.

(KI-generiertes Symbolbild: Gemini / AI Connect)

Financial institutions worldwide are accelerating the automation of core operations, yet basic safeguards are falling behind. A global survey released on November 12, 2026, by SAS, FT Longitude, and Microsoft titled 'Why AI at Scale Needs Governance at Speed' highlights a widening divide between adoption and oversight. The report surveyed 300 senior banking executives across 31 countries, delivering concrete metrics on the operational exposure across the international banking landscape.

According to the findings, 76 percent of surveyed banks project that artificial intelligence will autonomously decide on credit approvals and balance sheet risk positions by 2031. Current deployment is already widespread, with 82 percent of institutions utilizing generative AI, while 41 percent are experimenting with agentic systems. These autonomous agents operate beyond simple retrieval or drafting, executing functional processes without requiring continuous human approval.

Despite plans for widespread automation, operational risk controls remain underdeveloped. Only 18 percent of institutions have end-to-end AI governance frameworks in place. Furthermore, 42 percent of banks possess no emergency shutdown mechanism, often known as a kill switch, to immediately halt malfunctioning models during live operations. Only 19 percent of the surveyed firms maintain continuous, real-time monitoring over their deployed models.

This governance deficit emerges just as regulatory scrutiny over financial machine learning is beginning to solidify. On November 12, 2026, the Alabama State Banking Department issued a CEO letter announcing that it will incorporate the Conference of State Bank Supervisors supervisory framework directly into regular bank examinations starting in 2027. Regulators are moving to fill gaps left by federal model risk guidance, placing particular focus on third-party vendor oversight and model inventory verification.

At the same time, traditional banking groups are pushing back against uncontrolled external integrations. On November 14, 2026, the Bank Policy Institute issued an analysis warning of severe legal exposure tied to autonomous third-party agents operating within open banking environments. The trade group emphasized that unlike registered financial advisors, software agents carry no legal fiduciary duty, leading banks to reject liability for erroneous transactions or prompt-injection attacks.

Macroeconomic concerns are also mounting among central bankers. Speaking at the BlackRock Leader Circle in Frankfurt on November 12, 2026, Deutsche Bundesbank Vice President Sabine Mauderer cautioned that the substantial capital absorbed by AI data centers and chips is competing directly with sovereign debt and energy transition financing. Concurrently, a Federal Reserve Board working paper by Carollo et al. found that economy-wide productivity gains have yet to match the enormous capital expenditure driving computing infrastructure.

What this means for you

For banking professionals, AI governance is transforming from an abstract audit checklist into an urgent balance sheet defense. As state supervisors enforce new oversight frameworks from 2027, banks must urgently install technical kill switches and real-time monitoring to mitigate systemic failure risks.

Evidence

Solidly sourced
62/100
  • According to a study by SAS, FT Longitude, and Microsoft surveying 300 bank executives, 76 percent expect AI to independently decide credit and risk allocations by 2031.

    single source
  • The study found that 42 percent of banks lack an emergency kill switch for live AI models, and only 18 percent have comprehensive AI governance processes.

    single source
  • The Alabama State Banking Department announced that the CSBS AI framework will become mandatory in regular bank examinations starting in 2027.

    single source
  • The Bank Policy Institute warned on November 14, 2026, that third-party AI agents lack fiduciary duties toward account holders, raising legal liability concerns.

    single source

The evidence score is computed, not hand-set: from confidence, the number of sources and the share of verified statements.

Source & transparency

As of: November 15, 2026

AI-generatedAI-generated: produced automatically from vetted sources with technical quality checks (source, quote and figure verification); no human sign-off of each item before publication

Sources
3
Verified statements
0 / 4
Evidence score
62Solidly sourced

Want to put this into practice?

We connect you with suitable AI providers from the DACH region, free of charge and without obligation.

What's next?