Can Wall Street keep rogue AI out of its systems?

Bank of America, Citi, Goldman, JPMorgan and Morgan Stanley used earnings calls to detail AI cyber defenses

|
Published October 15, 2026
Can Wall Street keep rogue AI out of its systems?

Bank of America hired 1,000 outside contractors to patch software flaws exposed by powerful AI models, and they haven't left. CEO Brian Moynihan said so on Wednesday's earnings call. Wall Street's biggest lenders were all trying to convince investors that rogue AI is under control.

Moynihan stated to CNBC that the outside contractors have been brought in for the second quarter and are currently working alongside 3,000 in-house personnel.

This was referred to by him as "Mythos remediation" due to the fact that the Mythos model of Anthropic can spot potential bugs in software that could be used by attackers.

It's a very expensive bill. According to Moynihan, for the quarter the cost of remediation has come to "a couple hundred million dollars plus".

This was attributed to two different tasks: the Mythos one and the fixing of controls related to anti-money laundering at the bank. He said he expects costs from the former to go down.

How did Citi, Morgan Stanley, Goldman and JPMorgan respond?

Citigroup's Jane Fraser said on Tuesday that the bank doesn't wait for risks to happen. It has been hardening its systems and stress-testing its responses. Morgan Stanley's Ted Pick said the firm will spend "whatever it takes", because "you can't engage head-to-head with AI out of your garage."

The CEO of Goldman Sachs, David Solomon, mentioned "unchecked advancements in frontier models", while Jamie Dimon, the CEO of JPMorgan, was even more specific. He stated that he would not allow "just any outside agent" into his company, as the risk is too high for his clients.

According to Business Insider, this concern has been raised due to recent events. Thus, in July, an OpenAI agent supposedly breached Hugging Face AI's internal networks during a cybersecurity experiment.

After that, OpenAI agents were allegedly responsible for hacking an Australian government website. Then, the Prime Minister of Australia, Anthony Albanese, informed Sam Altman of their country's extreme concerns.

Anthropic made Mythos available in April to a small group of companies, banks and government agencies through Project Glasswing.

Pareesa Afreen
Pareesa Afreen is a reporter and sub editor specialising in technology coverage, with 3 years of experience. She reports on digital innovation, gadgets, and emerging tech trends while ensuring clarity and accuracy through her editorial role, delivering accessible and engaging stories for a fast-evolving digital audience.
Share this story: