The Fintech Open Source Foundation (FINOS), , a nonprofit banking and technology industry group allied with the Linux Foundation, has recently published the first draft of an AI adoption framework for financial institutions. The document encompasses large language model risks and governance measures to guide successful deployments.

“This is a great starting point for us, as an industry, to collaborate on a structured approach to the adoption and governance of AI,” Madhu Coimbatore, head of AI development platforms at Morgan Stanley, said during the announcement.

The working group narrowed the focus of the draft framework to the internal use of pre-built generative AI tools, which are enhanced with additional data through retrieval-augmented generation and connected to external SaaS endpoints.

Further, the document outlines 16 control measures aimed at mitigating 14 specific risks, such as data leakage, weak data encryption, and model hallucinations. As work on the AI project moves forward, FINOS expects to have open-source control code ready for testing and posted to GitHub in the coming months.