AI-found software flaws led to rushed patches behind NSA and bank outages
A wave of nationwide outages over the past week—including grounded flights, failed emails, and broken bank transactions—was not caused by a coordinated cyberattack, according to officials. Instead, AI appears to have played an unexpected role: advanced AI models are finding software flaws faster than companies can safely fix them, and rushed, poorly handled patches have been linked to some of the outages.
A spokesperson for Senator Mark Warner's team told Fortune that botched patch implementations were connected to an outage this week of an internal communications system at the National Security Agency (NSA), as well as the banking outages. Bank of America, Wells Fargo, and JPMorgan Chase all suffered outages this week.
Key points at a glance
- Officials say the recent outages were not a coordinated cyberattack.
- AI tools are finding software vulnerabilities at machine speed, overwhelming IT teams with fixes.
- A spokesperson for Senator Mark Warner's team linked botched patch implementation to the NSA internal communications outage and the banking outages.
- Bank of America disclosed last month it hired 1,000 contractors to patch flaws exposed by Anthropic's Mythos model, on top of 3,000 internal staffers, CNBC reported.
- JPMorgan Chase says AI tools were not a factor in its outage.
What lawmakers and officials are saying
Senator Warner said in a statement, "It feels like every day brings another reminder that AI is moving faster than our ability to manage its consequences." The Washington Sun separately reported that bank CEOs and the NSA told Congress that AI-driven patches were to blame for the service disruptions.
The patching problem, explained
A patch is a software update that fixes a security flaw. The irony, according to those following the story, is that tools meant to prevent outages may be contributing to them by creating new forms of human error.
Palo Alto Networks CEO Nikesh Arora said on X that "some of the recent outages are a consequence of a deluge of patches." He added that between teams' mandates to plug security holes and the threat of AI-assisted hacks, "we have a challenge" with "no silver bullets."
Cliff Steinhauer, Director of Information Security & Engagement at the National Cybersecurity Alliance, told Fortune that the volume of fixes is pressuring staff. "We don't always have time to fully test every update because there are so many and attackers can pick up the known vulnerabilities and exploit them," he said.
Experts disagree on where the blame lies
Amit Shah, founder and CEO of InstaLILY AI, called it "a failure of implementation." He told Fortune: "The AI finds a vulnerability in seconds and may not know what else breaks when you fix it. The person approving the patch sees a recommendation, not the systems it touches. So a tool built to prevent outages can end up causing them."
Jeff Collins, CEO of the IT platform WanAware, said anxiety around "the theoretical risk of a breach" is prompting a rush to eliminate vulnerabilities. "AI now discovers vulnerabilities at machine speed while our testing and deployment safeguards remain stuck at human speed," he said.
Steinhauer, however, does not see this as an implementation failure. He views it as a necessary phase as companies secure their code. "I think we are in a bell curve of vulnerabilities discovery and patching. The hope is that we'll get to a peak, then it will start to descend and slow down once we start to fix the underlying issues in our software," he said, comparing it to road maintenance: shutting a bridge to fix a crack is disruptive, but a collapse would be worse.
Who is responsible—human or AI?
Responsibility remains hard to pin down. A human may not have properly vetted the impact of fixes, but it is unclear what information they were working with from the AI tool. Open questions include whether the AI overstated the significance of vulnerabilities, whether it recommended changes in bulk that a human approved without proper vetting, and whether managers are pushing teams to adopt AI quickly to avoid a breach.
Collins argued the real failure is that "we plugged a hyper fast diagnostic engine into legacy operational pipelines never built to absorb that velocity of change."
The AI cybersecurity business behind the outages
OpenAI offers cybersecurity models that help organizations find vulnerabilities at scale. Its Daybreak program, which it calls an effort to "patch the planet," gives approved security professionals powerful defensive tools, including dashboards that classify bugs by severity. Anthropic's equivalent program is called Project Glasswing.
On Nov. 5, a week and a half before its outage, OpenAI president and co-founder Greg Brockman announced a partnership with JPMorgan Chase to use OpenAI models to help secure its systems. A JPMorgan Chase spokesperson said OpenAI's tools were not a factor in the outage, attributing it to "a temporary issue related to a software bug in on-premise hardware systems." The spokesperson added that the company did not identify that bug through AI tools or fix it with AI.
Brockman recently told The Information there was "no question" that cybersecurity would eventually become a bigger business for OpenAI than coding. The company replaced its Chief Revenue Officer earlier this year with Dali Rajic, tasking him with growing that business using his cybersecurity enterprise sales experience.
What is confirmed and what is still unclear
Confirmed: the outages were not a coordinated cyberattack, per officials; Bank of America, Wells Fargo, and JPMorgan Chase suffered outages this week; a Warner spokesperson linked botched patch implementations to the NSA and banking outages; Bank of America's contractor hiring was disclosed last month, per CNBC.
Still unclear: the exact role AI played in each individual outage. JPMorgan Chase disputes any AI connection in its case. It is also unresolved who bears responsibility when an AI-recommended patch causes problems—the human who approved it, the AI vendor, or the process itself.
Why this matters for banking and critical systems
The outages affected everyday services people rely on, including bank transactions and flights. The episode shows that AI-powered defensive security tools, widely adopted to prevent breaches, can introduce new risks when the fixes they generate are applied too quickly or without enough testing. Lawmakers are also responding: Representative Andy Ogle plans to introduce legislation today calling for better coordination as AI speeds up the discovery and exploitation of software flaws, NextGov reports.