Wall Street prepares for AI's "day after"

Customers providing login credentials to agents is leading to unvetted AI agent activity across America's largest banks, as reported by Axios on Thursday.

Financial institutions are now engaging in disaster scenario-planning alongside major AI firms, a move that signals the serious risk perception among large corporations.

This current activity differs from a summer incident involving a Hugging Face startup, where an AI agent swarm compromised its platform, sparking an industry security crisis.

Instead, the banks are concerned about the wider repercussions stemming from users permitting agents access to their accounts to take actions, particularly since these agents might act without user consent.

These preparations stem from worries about frontier AI systems, prompting banks to invest in defenses and plan for severe outcomes.

This report arrived amid mounting concerns over a growing cybersecurity crisis, highlighted by Salesforce CEO Marc Benioff posting "code red" on X.

An Anthropic security employee, quoted by Benioff, stated, "as bad as things have gotten in the past 4 days, they are going to get a lot worse before they get better."

This comment related to researchers discovering major vulnerabilities in a programming language powering millions of websites, marking a significant AI-related security finding.

JPMorgan CEO Jamie Dimon, speaking in Chicago this month, cautioned about cyber threats or AI agents taking unapproved steps, stating, "You're assuming they're going to act properly on your behalf and they may not."

Dimon insisted that stopping AI is impossible; "You can't stop it. You just have to, like, manage the hell out of it."

Banks are actively blocking agent access, though they are also investigating secure methods to let customers connect with desired services.

A JPMorgan spokesperson told Axios, "Under no circumstance should customers share passwords with any person or thing. Period," adding that secure open banking platforms exist.