Financial companies are finding ways to let employees use AI tools without opening the door to security risks. Two firms, MoonPay and Equals, shared how they are handling this balance.
MoonPay is a cryptocurrency payments company based in New York. Doug Innocenti serves as both the chief information officer and chief information security officer there.
Innocenti said his goal is to avoid becoming the "party of no" when it comes to AI. Instead, his team looks for ways to make new tools safe to use.
Managing Shadow AI From ChatGPT to Agents
When ChatGPT first launched in late 2022, MoonPay worked with a startup called Aim Security. At the time, there was no enterprise version of ChatGPT with built in protections.
Aim Security built a secure portal that added anonymization and blocking features ChatGPT lacked. As more AI tools appeared, Aim expanded to scan internal tools for hidden AI components.
This let MoonPay check which AI models a tool used and how it might use company data. Cato Networks bought Aim Security in 2025 and renamed it Cato AI Security.
MoonPay is now looking at more tools from Cato's network security lineup. Innocenti said he wants AI defense to work at the network level, not just on individual computers.
As MoonPay starts using AI agents like Claude Cowork, Cato AI Security helps watch for sensitive data being shared with these tools. Innocenti said agentic AI changes so fast that he researches new developments constantly.
Earlier this year, Innocenti took on the CISO title in addition to his CIO role. He said this merger has helped IT and security staff work together instead of competing.
Employees now feel comfortable asking questions before using new AI tools, he said. This has made staff a stronger part of the company's security approach rather than a weak point.
New Tools Address Growing AI Threats
Equals is a payments platform based in London. The company said 86% of its application code is now written by AI agents, up from 5 to 10% a year ago.
James Simcox, the company's chief operations and product officer, said AI agents have sped up legal reviews and customer support work. But he also pointed to risks from employees adjusting security settings on their own.
Simcox said staff sometimes give AI tools more access than needed just to get their jobs done. He said this isn't done with bad intent, but it can still expose company data.
No employees have lost jobs to AI at Equals, which reported double digit revenue growth and more than £58 billion in transaction volume over the past year. The company has shifted developer focus toward more technical work while doubling its security team size.
Equals replaced several separate security tools with one platform called Wiz. It also added Okta tools to track staff and customer identity activity across systems.
Simcox said these tools help catch behavior that looks unusual, like a single login used many times in seconds. The company also monitors suspicious transactions rather than blocking every unusual login outright.
Simcox said the shift toward AI agents has made cyberattacks look more like human activity, making them harder to catch. He said the company's security approach depends on the type of business and its software habits.
Both companies said the key is building processes and tools that support safe AI use rather than shutting it down entirely.
