‘Twas the week of big bank earnings, and Wall Street had a new threat on its mind: rogue AI.
Executives at the country’s largest financial institutions used their earnings calls to assure investors that they are doing everything in their power to fend off attacks from powerful, rogue AI models.
Some were specific about the measures they were taking, like Bank of America’s CEO, Brian Moynihan.
Moynihan said during a Wednesday earnings call and an interview with CNBC that the bank had rushed to shore up vulnerabilities in its codebases after the release of powerful AI models such as Anthropic’s Mythos.
The firm hired 1,000 external consultants to patch these issues over the last quarter, and their work is still continuing.
Citigroup CEO Jane Fraser shared similar sentiments on AI safety during the company’s third-quarter earnings call on Tuesday.
“We don’t wait for risks to happen,” she said, adding, “We’ve been hardening our systems and running challenging and disruptive scenarios to stress test our defenses and our responses.”
Later in the call, she said that Citigroup was continuing to “invest in leading-edge capabilities” to strengthen its cyber defenses.
Morgan Stanley CEO Ted Pick said in a Wednesday call that the firm is investing “whatever it takes” to protect its clients’ assets, because “you can’t engage head-to-head with AI out of your garage.”
Goldman Sachs CEO David Solomon said the bank sees concerns about the “unchecked advancement of frontier models” and that AI presents significant risks alongside opportunities.
“We must enable innovation and also demand responsibility,” Solomon added in the Tuesday call.
And finally, JPMorgan CEO Jamie Dimon said in a Tuesday call that he wouldn’t allow “just any outside agent to come inside the company, because that’s just too risky for clients and for JPMorgan.”
Rogue AI agents are top of mind
The comments from the finance executives come as rogue AI agents by labs like OpenAI, Anthropic, Google, and Meta, capable of exploiting software vulnerabilities, have hacked into the software and website of several organizations in recent months.
One of the first instances of this occurred in July, when an OpenAI agent hacked into the internal systems of AI firm Hugging Face during a cybersecurity test.
More recently, its agents hacked into Australia’s government website, leading Prime Minister Anthony Albanese to express Australia’s “extreme concern” about the incident to OpenAI CEO Sam Altman.
AI staffers themselves have become increasingly concerned about the pace of AI development and the risks it poses. In September, Anthropic researcher Jacob Coxon publicly resigned, saying that Anthropic and OpenAI were “gambling with lives,” prompting many other employees at frontier labs to chime in with similar concerns.
Shortly after, chiefs of AI companies, such as Altman, Dario Amodei, and SpaceXAI CEO Elon Musk, all agreed on “pacing the frontier,” referring to slowing AI development and adding safety checks and balances.
