Technology Miraj Islam October 16, 2026 0 Comments

Bank of America had to hire 1,000 contractors to fix software flaws detected by Anthropic Mythos

Bank of America hired 1,000 contractors and spent hundreds of millions to fix vulnerabilities discovered by AI
Anthropic’s Mythos Uncovered Enough Flaws to Trigger Large-Scale Remediation at Major Banks
Financial institutions warn that advanced AI security tools are exposing weaknesses faster than teams can fix them

Bank of America, one of the largest financial institutions in the United States, was forced to hire a thousand people in contractor roles and spend more than $100 million to fix all the bugs found by the famous Mythos, Anthropic’s generative artificial intelligence (GenAI) model.

When Anthropic first introduced Mythos in April 2026, it described it as a tool with advanced cybersecurity capabilities that can autonomously discover and exploit previously unknown software vulnerabilities.

However, the tool was so capable (and dangerous) that Anthropic blocked it, giving access to a handful of vetted organizations: the US government, critical infrastructure organizations, banks, and some cybersecurity companies, to name a few.

Bank of America and JPMorgan Chase

Bank of America was apparently one of those companies, as he spoke with CNBCThe bank’s CEO, Brian Moynihan, said that when Mythos started finding bugs in the system, the company saw it as a “real risk” and was forced to sit down and come up with a plan to address all the vulnerabilities.

“We hired 1,000 contractors to help us do the patching work in the second quarter, and those people remain here today, helping us, in addition to the 3,000 people we have in that area,” he told the publication. CNBC He added that Moynihan told analysts earlier in the day that the company spent “more than a couple hundred million” on fixing bugs and addressing regulatory issues.

Even for a bank of that scale, such hires and expenses are rare, to put it mildly. That’s probably why Moynihan added that the increase in spending was temporary and that once the vulnerabilities are addressed, there won’t be as much work as there is now. “Once the patch is applied, you don’t need to apply as much patch…The pipeline moves forward,” Moynihan said.

In his report, The next website Put it in the context of complying with state and federal regulations. “The compliance spending follows Treasury sanctions of more than $1.21 billion announced on October 13,” the publication wrote. “Federal regulators said the bank had failed to report $5 billion worth of suspicious transactions and complete mandated steps in 2024.”

serious consequences

Bank of America isn’t the only bank rushing to fix what Mythos exposed. A few months ago, JPMorgan Chase CEO Jamie Dimon said the bank was working with peers and vendors to make sure its systems are prepared for what’s coming. “It’s serious work,” Dimon said in a Bloomberg TV interview. “I think we now have hundreds of people doing it full time.”

Goldman Sachs CEO David Solomon said in mid-April that he was “very aware” of Mythos’ capabilities, adding that the firm was working “closely” with the technology company. “Obviously, LLMs are progressing rapidly and we are very aware of the enhanced capabilities of these new models with the help of the US government and model publishers,” Solomon told analysts at the time.

When Anthropic first revealed Mythos, it said AI models have reached a level of coding capability “where they can outperform all but the most skilled humans at finding and exploiting software vulnerabilities.”

“The consequences – for economies, public safety and national security – could be serious.”

Many companies that have tried it seem to agree. Mozilla, for example, called it “as capable as the best security researchers in the world,” after finding 22 security-sensitive bugs in just one round of testing.

Palo Alto Networks, on the other hand, used Mythos to find more than two dozen critical vulnerabilities in about three weeks, but testing also showed that companies needed significantly larger AI security budgets.