State of play: Banks are facing unchecked AI agent activity, driven primarily by customers providing login credentials to agents, who in some cases appear to have taken action independently of user prompts, Axios reported Thursday.
- This activity is distinct from the malicious behavior involved in a summer attack on a startup called Hugging Face, where a swarm of AI agents hacked its platform in an episode that sparked an industry security crisis.
- Instead, banks fear the ripple effects and risks of users allowing agents to log into their accounts and take action, especially since agents might take actions that users have not sanctioned.
- Concerns about border AI systems, not AI agents, have led banks to invest in prevention and prepare for the worst.
Zoom: The Axios report comes amid growing concerns about a burgeoning cybersecurity crisis.
- Salesforce CEO Marc Benioff issued a “code red” on
- The Anthropic employee was referring to security researchers who discovered major vulnerabilities in a programming language used to run millions of websites, marking one of the most significant cybersecurity discoveries ever made with AI.
- The Axios report on banks, which referred to agentic AI and not frontier AI systems, sparked calls for regulation from across the political spectrum, including from Sen. Bernie Sanders (I-Vt.), Florida Gov. Ron DeSantis and Sen. Mark Warner (D-Va.).
- Former White House AI chief David Sacks said banks should support user-directed agents that allow users to “grant, limit and revoke access,” with read-only as an “intermediate step.”
- Sacks said it’s irresponsible to conflate “user-authorized agent behavior with cyberattacks.”
Flashback: Major banks participated in early testing of Anthropic’s Mythos, the AI company’s most cyber-capable model.
- It was part of an effort by AI companies and big banks to use the technology to better understand and prepare for cyberattacks or other risks.
- Since then, banks have been working on plans to not only avoid AI-related issues, but also to protect accounts, reassure customers and keep operations running in the event of a large-scale attack.
Reality check: It’s not just a standard, it’s an obligation for critical industries that must meet the needs of government, shareholders and customers.
- Scenario planning is a requirement for critical infrastructure providers, including in the financial sector. Avoiding it would be professional misconduct.
- The riskiest scenarios involve fear: During a speech in Chicago earlier this month, JPMorgan CEO Jamie Dimon warned of the potential for cyberattacks, or the possibility of AI agents taking actions that users have not approved.
- “You assume they’re going to do the right thing on your behalf and that may not be the case,” Dimon said. The only answer is vigilance, he said, warning that it would not be possible to stop AI. “You can’t stop it. You just have to get through it.”
- Last month, an economist raised the possibility of a bank run caused by agents moving their users’ household cash into higher-yielding accounts.
Between the lines: Banks want their customers to know that they have not seen any malware activity or evidence of a hack affecting customer funds or accounts.
- Banks strongly advise users not to provide their passwords to agents, especially since services exist or are being developed that allow customers to have read-only access. The startup Plaid is one such platform that can, among other functions, extract balances and transaction history.
- “Under no circumstances should customers share their passwords with anyone. Period,” a JPMorgan spokesperson told Axios. “There are open and secure banking platforms that do not involve handing over your account keys. Customers are the first line of defense in protecting their accounts.”
- JPMorgan has taken steps to block agent access, as well as other banks, even as it explores secure ways to allow customers to connect to the services of their choice.
The plot: AI companies expect political backlash. Banks are preparing for possible operational disruptions.
- The extent of federal planning for a crisis spanning both remains unclear.
The bottom line: The pace of advances in AI has Wall Street and Silicon Valley preparing for worst-case scenarios, while Washington works to mitigate cyber risks and relies on voluntary security commitments.
Gn bussni
