Breaking

Bank of America's Moynihan Joins Wall Street Chorus Warning Over Anthropic's Cyber-Capable 'Mythos' AI

July 16, 2026 at 14:53 EDT

Security
Foundation Models
Finance

Bank of America Chief Executive Brian Moynihan added his voice on July 16, 2026, to a growing group of Wall Street leaders sounding the alarm over Anthropic's cyber-capable artificial-intelligence model, Mythos, calling it a serious issue that banks are working hard to address. Speaking on Bloomberg Surveillance, Moynihan urged reassurance that the risk is being taken seriously while emphasizing the speed at which such systems can affect vulnerabilities and the corresponding need for rapid response.

July 16, 2026 · Wall Street

Bank CEOs Sound the Alarm on Anthropic's Cyber AI, "Mythos"

BofA chief Brian Moynihan joins JPMorgan, Citigroup and Goldman Sachs leaders warning about Claude Mythos — a frontier model with state-of-the-art cyberattack capabilities. Released only to select banks for defense, its speed is the central concern as lenders race to patch vulnerabilities.

3/10

complex cyber tasks solved end-to-end in independent evaluation

~22

average steps to autonomously complete a multi-stage attack

April '26

emergency briefing convened by Treasury & the Fed for bank CEOs

Two models, two access levels

Claude Fable 5

Safeguarded public counterpart — open general release

Claude Mythos

Restricted preview — select banks & orgs only, for defensive use

Higher stack = broader access. Mythos was withheld from public release because of offensive misuse risk.

Defensive value

Used by banks for internal vulnerability scanning
Enables faster patching and info-sharing
Project Glasswing steers usage toward protection

Misuse concerns

Finds & exploits OS and browser weaknesses
Dimon: access would be like "ballistic missiles"
Some reproduced findings with public models

The core worry

"Not a sentient super-hacker — but a powerful engine for exploit chaining that can act at machine speed."

Fully automated attack chains stay limited for now, but the pace at which Mythos can operate keeps regulators and lenders on edge — leaving the defense-vs-misuse debate far from settled.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.