The bank brought in outside technical workers after Anthropic’s Mythos tool uncovered software vulnerabilities across its systems.
Bank of America brought in an additional 1,000 contractors during the second quarter to help repair software vulnerabilities discovered by an artificial intelligence model, Chief Executive Brian Moynihan announced on Wednesday. Speaking during the company’s third-quarter earnings call and in an interview with CNBC, Moynihan said the outside workers were recruited after Anthropic’s Mythos model began identifying security weaknesses in the bank’s internal systems.
The contractors joined 3,000 existing technical personnel at the bank and remain on the job today, TechRadar reported. Moynihan said the bank spent “a couple of hundred million plus” on remediation work that combined fixing the software flaws with meeting recent regulatory mandates.
Software flaws and surging remediation costs
Anthropic introduced Mythos in April 2026, describing it as an advanced system capable of autonomously finding and exploiting unknown software flaws. Because of the risks tied to those capabilities, Anthropic limited access to vetted groups, including the United States government, critical infrastructure operators, cybersecurity firms and major financial institutions.
Bank of America treated the findings as a genuine risk that required an immediate response plan. Moynihan described the hiring and spending surge as temporary, telling analysts that the remediation burden should decline once the initial wave of patches is completed. According to The Register, Moynihan did not specify how much of the remediation spending went directly toward Mythos patches versus compliance work tied to recent regulatory orders.
Industry reaction across Wall Street and tech
Other major financial institutions and technology firms have also dedicated staff to address vulnerabilities flagged by the model. JPMorgan Chase Chief Executive Jamie Dimon said in an interview with Bloomberg TV that his bank has hundreds of people working full time on the issue with peers and vendors, adding, “It’s serious work.”
Goldman Sachs Chief Executive David Solomon told analysts in mid-April that he was hyper-aware of Mythos’s capabilities and confirmed the firm was cooperating closely with Anthropic and government officials.
Outside the banking sector, Mozilla tested the tool on Firefox 150, identifying 271 flaws. Mozilla technology head Bobby Holley stated that the system showed capabilities matching elite human security researchers. Palo Alto Networks separately used Mythos to locate more than two dozen critical flaws over roughly three weeks, later noting that such automated testing demands significantly larger security budgets.
Regulatory fines and workforce shifts
The software patching push took place alongside significant regulatory penalties and internal restructuring at Bank of America. The bank’s shares dropped more than 8 percent on Wednesday morning after it reported a 17 percent decline in quarterly profits.
That earnings decline followed a $1.21 billion settlement announced on 13 October with the Treasury Department’s Financial Crimes Enforcement Network. The agency found that the bank failed to report more than $5 billion in suspicious transactions, including Chinese money laundering, leaving the firm fined $1.2 billion for money laundering lapses.
Regulators also cited Bank of America for failing to finish corrective steps ordered in 2024. The Office of the Comptroller of the Currency assessed a separate $750 million penalty for Bank Secrecy Act and anti-money laundering deficiencies, while the Office of Foreign Assets Control imposed an $11.6 million penalty over sanctions violations tied to Iran, Russia, narcotics and terrorism. During the same period, the bank cut approximately 3,000 roles, including a 5 percent reduction in managers, though it did not specify whether automation drove those job cuts.
Projected returns from AI projects
Despite the remediation expenses, Moynihan told analysts that internal artificial intelligence initiatives have increased the bank’s technical engineering efficiency by about 10 percent. The company currently operates roughly 130 AI projects, each anticipated to generate a two- to three-fold return on investment.
Moynihan said the bank expects these projects to produce between $250 million and $300 million in benefits during 2026. The bank projects those financial benefits will rise to $700 million by the end of 2027.
