Bank of America CEO Moynihan Warns of AI Model Security Risks
0xBroomberg
Bank of America CEO Brian Moynihan says Anthropic's AI model Mythos is raising serious security concerns across Wall Street, with major banks racing to patch vulnerabilities; meanwhile, BofA's investment-banking revenue jumped 47% year-on-year to $2.2 billion.
What is Mythos, and why is Wall Street nervous?
Mythos is an AI model released by Anthropic earlier this year, designed to find security vulnerabilities in systems — in plain terms = it can automatically spot weak points in a bank's IT infrastructure.
The model is not publicly available; only select Wall Street firms have access for internal testing, and Bank of America is one of them.
This means → once this capability spreads more widely, attackers could find flaws far faster than banks can fix them.
What are the two banking chiefs saying?
Moynihan said Mythos has "significantly increased the workload" — the speed at which vulnerabilities surface, and must be patched, has changed dramatically.
JPMorgan CEO Jamie Dimon was blunter: giving broad public access to Mythos would be like "handing ballistic missiles to individuals."
This reflects a shift on Wall Street from "worth watching" to "act now" — when the heads of the two largest U.S. banks speak out in the same week, the urgency is real.
How is Bank of America responding?
Moynihan said BofA is using Mythos to test its own systems and sharing findings with peer institutions.
The bank is also ensuring third-party vendors meet their security obligations, with teams collaborating internally.
This means → the defense logic is "use AI to find your own holes before an attacker does" — essentially a speed race.
How are the earnings holding up?
BofA's second-quarter results were strong: investment-banking revenue rose 47% year-on-year to $2.2 billion, driven by trading and advisory.
Moynihan described the M&A pipeline as "robust" — deals in mergers, strategy, and financing are being announced daily.
In plain terms = security risk is one headline, but the money-making machine hasn't stopped — banks are patching vulnerabilities and posting record numbers at the same time.
What comes next?
The core test is balance: banks need AI tools to stay competitive, yet must guard against the systemic vulnerabilities those same tools create.
As AI models grow more capable, pressure on both the financial industry and the U.S. government to assess threats is rising in parallel.
This means → AI safety is no longer just a tech-company issue — it is becoming a central question for financial regulators. The next thing to watch is whether policymakers move to restrict access to models like Mythos.
Content is for reference only, not financial advice.