Banking's AI Problem Isn't Adoption. It's Visibility.
Security Boulevard, Wednesday, June 10th, 2026
Untracked, ungoverned AI use across banks creates data, model-risk, and regulatory exposure that risk teams can't see.
The article argues that banking's AI challenge is not adoption but visibility, as AI use spreads from the trading desk to the back office through unapproved tools that risk and IT teams cannot see.
It cites real-world examples: credit analysts pasting loan files with personal data into ChatGPT, relationship managers using summarizers on confidential portfolios, traders using free LLMs with material non-public information, and back-office staff using Copilot on account numbers, none appearing in audit trails.
Industry-wide, 90% of AI usage is untracked and ungoverned, and 97% of generative AI users have already faced linked security incidents. For banks, untracked tools risk GLBA Safeguards Rule breaches, SR 11-7 model risk violations, and NYDFS Part 500 and third-party risk failures. Syndicated from the FireTail blog.