Enterprise & Security

Bank of America: 1,000 Contractors Patching Flaws Claude Mythos Exposed

2 min read AI-generated

The contractors came on in the second quarter, on top of 3,000 internal staff, and they are still there. Together with anti-money-laundering work, that cost a couple hundred million dollars in Q3.

Featured image for "Bank of America: 1,000 Contractors Patching Flaws Claude Mythos Exposed"

Bank of America CEO Brian Moynihan told CNBC on Wednesday how his bank responded to Claude Mythos: with 1,000 outside contractors who have been patching software since the second quarter, alongside 3,000 internal staff. The trigger is that frontier models surface vulnerabilities in proprietary and vendor code, some of it decades old.

“That risk is real risk, and that’s why when the frontier-type models came out. … A lot of us sat down and said, ‘How do we figure out how to get through this quickly?’” Moynihan said in an interview after the bank’s quarterly results.

What it cost

A “couple hundred million plus” is the figure Moynihan gave analysts for software remediation together with anti-money-laundering compliance in the third quarter. The stock fell as much as 8% the same day.

He frames it as temporary all the same: “Once you patch, you don’t have to patch as much. … The pipeline moves through.” Remediation costs should decline over time.

Per CNBC, these are among the first concrete numbers any large US corporation has given on the manpower and money spent defending against Mythos. The model went out in April to a small group of banks, technology firms and government agencies.

The other side of the ledger

The same bank runs about 130 internal AI projects. Savings from them are projected to rise from $300 million this year to $700 million in 2027. Headcount fell by roughly 3,400 over the past year, with about 5% of managers cut; how much of that is attributable to AI is unclear.

Moynihan left one line worth reading twice: “We’ve got to stay ahead of them, and we’ve got to be mindful that the people doing the software development are the people trying to sell us the services.”

The same capability, billed twice

What Anthropic sells as a security capability shows up on a customer’s books as a cost center, and it does so before that customer deploys the model at all. Mythos finds holes in code that ran quietly for decades, and creates work nobody had budgeted for.

For the rest of the market, that’s the interesting number. Not what Mythos access costs, but what it costs to react to a model you never licensed. Bank of America hired 1,000 people for it. Smaller organizations don’t have that option.

Sources

Bank of AmericaClaude MythosSecurityEnterpriseCosts