Banks Push for Clarity on Outsourced Models

Third-party models are now part of everyday banking—but they come with challenges banks can’t ignore. According to ProSight’s 2025 Model Risk Management Survey, 70% of banks use third parties for model development, implementation, or validation. And while outsourcing saves time and adds sophistication, many banks are frustrated by poor documentation and limited visibility into how these “black box” models actually work. 

Transparency Still Lacking 

Banks say it’s tough to get usable disclosures from vendors. Just 38% of respondents rated their vendors’ technical descriptions as moderately good or better. One risk leader described a typical experience: “When we ask for any type of technical documentation, performance, or even an internal validation report with redacted information, that's where it usually comes to a complete stop with a lot of them.” 

The stakes are high. In one case, a risk manager said his team discovered a serious flaw in an outsourced origination model. “[They had] a lot of fancy code and data processing, but they did not understand bank regulation,” he said. The bank declined to use the model. 

Stronger Contracts, Better Results 

Survey data shows that banks with stronger contract language are nearly three times as likely to get useful disclosures. But drafting airtight agreements isn’t always easy—especially when vendors assert that their models are proprietary. As one risk VP put it, “There's normally a middle ground, or if they have a lion's share of the market, they may just say no.” 

AI Adoption Is Accelerating 

Ninety-one percent of banks now use AI—most commonly in fraud detection, document processing, and security. But managing AI risk requires constant attention. “A machine learning model may change several times in one day,” one respondent warned. That’s why he favors co-ownership of AI governance between model risk and IT risk teams. 

Bottom line: Banks need better vendor collaboration and robust oversight if they want to unlock the benefits of third-party models and AI—without taking on more risk than they bargained for. 

For more, read “Proposed Contract Language Fills Gaps in Model Disclosure Standards.”