Artificial intelligence is moving beyond simply answering questions. New AI agents can act autonomously to complete tasks, and as the technology becomes more powerful, cybersecurity experts are raising questions about what could happen when those agents have access to your money.
“Normally we deploy an AI agent to perform some task autonomously. That autonomy is the core defining feature of what makes an agent an agent,” said Jake Williams, a faculty analyst at IANS.
But that independence can also create risk. Williams says a “rogue” agent doesn’t necessarily have to be malicious. It could still be carrying out the task it was given, just in a way its creators never anticipated.
“The idea that we could have a rogue agent is basically the agent operating against a task it was given, but in a way that was unanticipated... by the developer or the deployer,” Williams said.
The stakes become particularly high when an AI agent has access to someone’s finances.
“The decision-making which used to rest with us, meaning as humans, is now being shifted to a computer whose making decisions on your behalf,” said T.J. Marlin, CEO of Guardrail Technologies. “What happens if you ask an agent and give them your credentials to your bank account, and all of a sudden they transfer $100,000 to somebody that you didn't want them to?”
RELATED STORY | AI is changing how consumers shop, but bad recommendations could cost brands
That’s why experts say safeguards become increasingly important as AI is given more authority over sensitive information and financial accounts.
“If an AI is going to touch your money, it needs clear limits. You need a way to stop it, and you need a record of what it did,” Marlin said.
But as the threat evolves, banks are also turning to AI to fight AI. Jonathan Gould, the Comptroller of the Currency, said financial institutions are working to stay ahead of rogue AI agents and the potential risks they pose.
RELATED STORY | 43% of voters say they’ve been fooled by AI ads this campaign
“Banks are focused on defending the frontier and some of the most fortified parts of our system,” Gould said. He said banks have spent months “fortifying their defenses and using AI as a defense against the AI agents.”
As AI agents become more capable, the question for consumers may increasingly shift from what AI can do to how much control they’re comfortable giving it.
