Banking and tech nonprofit unveils draft framework for genAI adoption

The news: The Fintech Open Source Foundation (FINOS) released a draft of its generative AI (genAI) adoption framework for financial institutions (FIs), per CIO Dive.

• The group’s members include Nvidia, Moody’s, JPMorgan, Citibank, Goldman Sachs, and Morgan Stanley.

How we got here: Regulators have had a difficult time keeping up with this technology, which makes it risky territory for FIs. 

• President Joe Biden issued an executive order last fall that underscored the urgent need to oversee AI development responsibly and safely, emphasizing a unified federal approach due to the rapid advancements in AI capabilities.
• Multiple regulators around the world have confirmed banks are responsible for any AI-driven mistakes, but clear frameworks outlining best practices have yet to be established.

Dive deeper: The FINOS framework seeks to provide guidelines that could help FIs accelerate their use of AI technology. It recommends:

• Developing governance and control measures: The framework recommends FIs adopt 16 control procedures to address 14 specific threats, including data leakage, model hallucinations, and weak data encryption.
• Monitoring and auditing AI use: FIs should set up continuous monitoring of AI inputs and outputs, tracking potential security issues like data breaches or model inconsistencies. They should also implement logging and alert systems to track performance, security breaches, and regulatory compliance.
• Prioritizing human oversight: Even with advanced AI tools, humans must remain in charge of the technology’s inputs and outputs. This will help to identify risks like hallucinations or bias. 
• Focusing on compliance and security: The framework advises FIs to set up robust internal controls, such as strong encryption and access restrictions, to mitigate risks associated with AI usage.

Key takeaways: Fintech Business Weekly writer Jason Mikula said he doesn’t believe most financial services companies are well-positioned to “act on the wave of AI-related regulation … that is coming.” 

To better prepare, FIs should aim to implement safeguards as soon as possible. While not yet mandatory,  these standards are likely to influence federal banking regulators’ research.

Moreover, adhering to these guidelines can help FIs prevent genAI mistakes and avoid potential legal and reputational consequences.