HomeEverythingEducationTV
Equities & FundsCrypto & Digital AssetsAI & TechnologyBusiness & CorporateUS Politics & PolicyGeopolitics & Global RiskMacro, Rates & FXCommodities & EnergyEuropean Politics & MarketsAsia-PacificReal Estate & Property
Story archiveAll categories
← All Stories

Banks Automating GenAI Testing, But Scope Varies

Created at 17 Jul · 3:36 AM1 source↑ Market-relevant
IN SHORT

Nearly one-third of banks have begun automating generative AI model testing, according to a Risk Benchmarking study. However, the application of LLM-as-judge for autonomous sign-off remains limited, with most institutions using it for scaled testing rather than full automation.

✉Newsletter

PiQ Daily

Pick your topics. Get only what matters, on your cadence.

Key Numbers

one-thirdbanks partially automating GenAI testing

Who's Involved

Risk Benchmarking
Provider of the Model Risk study on GenAI testing

↳ Why This Matters

The adoption of GenAI in banking introduces new risks that require robust testing and validation. While automation is increasing, the varied scope of its application suggests that many institutions are still navigating the complexities of ensuring AI model safety, ethics, and reliability before full integration.

Key facts

  • Almost one-third of banks have partially automated the testing of generative artificial intelligence (GenAI) models.
  • The use cases for automated GenAI testing vary widely among financial institutions.
  • LLM-as-judge technology offers model testing at scale but is not widely used for autonomous sign-off by lenders.

A significant portion of banks are moving towards automating the testing of generative artificial intelligence (GenAI) models, with nearly one-third having already implemented partial automation. This trend is highlighted in Risk Benchmarking’s Model Risk study. Despite the advancements in LLM-as-judge technology, which enables model testing at scale, its application for facilitating autonomous sign-off by lenders remains limited. Most financial institutions are currently leveraging this technology for scaled testing rather than for fully automated approval processes.

Frequently asked questions

LLM-as-judge refers to the use of large language models to evaluate and test other AI models, offering a scalable approach to model assessment.

The primary finding is that while nearly a third of banks are automating GenAI testing, most are using it for scaled testing rather than for autonomous sign-off.

The information comes from Risk Benchmarking’s Model Risk study.

Get the newsletter.

Pick the topics you actually care about. We'll email when there's news worth your time, on the cadence you choose. Cancel any time from your account.

Cadence

How It Developed

Nearly one-third of banks have partially automated generative AI model testing.
Most banks are using LLM-as-judge for scaled testing, not autonomous sign-off.

Sources

T1
Banks are automating GenAI testing, but scope varies widelyRisk.net

Related Stories

Visa, Artemis: AI agent economy needs new payment infrastructure
16 Jul · 10:41 AM
Meta Oversight Board: Top AI models avoid criticizing repressive regimes
16 Jul · 10:03 AM
Carrington Mortgage Services partners with Kastle for AI agents
16 Jul · 6:11 PM
AI chatbots risk spreading government speech restrictions globally, study finds
16 Jul · 11:56 AM
Hong Kong AI education plan needs more resources, experts say
16 Jul · 9:36 PM