top of page

Report: Botched Software Upgrades Driven by AI Are the Cause of Tech Outages, Not Cyberattacks

Writer: WGON
WGON
3 hours ago
2 min read

A string of major outages across banking, air travel, and tech this past week sent anxious users searching for someone to blame, but the companies involved say the culprit was bad software updates, not hackers. In many cases, rushing to plug holes discovered by AI constributed to the failures.


The Wall Street Journal reports that the trouble started last Thursday, when more than 450 million Microsoft 365 users found themselves unable to send or receive email. By Friday the 13th, air travel across the country had ground to a halt after the FAA issued a nationwide ground stop, and the HR software company Workday went down too. The disruptions kept coming Monday, when Chase banking services stopped working and United Airlines lost the tools it uses to communicate with pilots. On Tuesday, the National Security Agency’s internal network crashed, and Apple suffered its own outage.


“The last 36 hours has been hysteria,” said Doug Madory, head of internet analysis at the technology company Infoblox, on Tuesday.


Every one of those organizations (Microsoft, the FAA, Chase, the NSA, Workday, Apple and United among them) has said the outages were not the result of hacking. In most cases, they point to a system upgrade gone wrong. Some of these upgrades are related to a rush to fix vulnerabilities located by AI.


Iain Mulholland, Salesforce’s chief information security officer and a former British Army officer, said a friend asked him on Tuesday whether she needed to change all her passwords. His advice: “calm down and carry on.” Mulholland said the instinct to suspect a cyberattack isn’t surprising given the current climate. “There’s been so much reporting of the impending AI apocalypse, and then when ordinary people see systems going offline, they immediately jump to the conclusion that it’s cyber, even if it’s not,” he said.


That climate has been building for months. About two months ago, tech executives warned of what some called an AI apocalypse, including Dario Amodei, head of the AI company Anthropic, who cautioned that rogue AI systems could take down the internet.

Cybersecurity commentators have taken to calling the past month “Black October,” pointing to what they describe as an unprecedented wave of intrusions, many of them fueled by a new AI hacking tool called Artex. Last month refrigerators across Europe were shut down in another Iran-linked incident.


 
 
 

Comments


bottom of page