U.S. tech outages expose operational risks around AI-era software changes

U.S. tech outages expose operational risks around AI-era software changes
AI-era software risks exposed

A string of service disruptions across U.S. banking, aviation and workplace software is intensifying scrutiny of how companies manage fragile digital infrastructure. While the incidents do not appear to be connected or caused by hacking, experts say the broader push to update aging systems for the AI era can still increase the risk of outages and cascading failures.

Highlights

  • Chase, United Airlines, and Microsoft experienced outages blamed on internal software bugs, technology problems, and capacity issues amid rising user demand.
  • Experts cite accelerated software changes and modernization pressures driven by AI, creating greater operational risks and straining review processes for reliability and safety.
  • Downdetector tallied nearly 6 million U.S. problem reports in the week ending Nov. 16, marking a one-year high, though monthly major outages remain below average.

Outages highlight strain on legacy systems

As first reported by Axios, Americans over the past week are struggling at times to access JPMorgan Chase banking services, board United Airlines flights and use Microsoft Outlook, as separate technical failures hit major platforms.

Homeland Security Secretary Markwayne Mullin suggests AI systems that can write and change code could be contributing to the disruptions. But experts cited by Axios say there is no evidence that AI-assisted hacking is behind the incidents, and the available explanations instead point to software patching, network surges and older equipment failures.

Chase blames an internal software bug that delays some transactions involving other banks. United cites a technology problem, while the FAA says a flood of network traffic destabilizes older legacy equipment after a recent nationwide ground stop. Microsoft says a months-old bug, increased traffic and reduced server capacity leave users unable to access email and other Microsoft 365 services.

Industry faces faster coding, bigger resilience risks

Experts say AI may still be influencing the operating environment indirectly, as institutions rush to modernize code and harden systems against vulnerabilities that advanced AI models can identify and exploit more quickly. That acceleration is increasing the volume of software changes and putting more pressure on review processes meant to catch reliability and safety issues before deployment.

David Woods, professor emeritus of integrated systems engineering at Ohio State University, says increasingly complex systems are being pushed to run faster and cheaper without enough attention to failure points. Bruce Schneier, a security technologist, adds that people often interpret coincidence as conspiracy, while Honeycomb technical fellow Liz Fong-Jones says the recent incidents have little in common beyond being examples of long-standing cascading failures.

Downdetector receives nearly 6 million U.S. problem reports in the seven days ending Nov. 16, the highest weekly total in a year, according to data cited by Axios from Ookla analyst Luke Kehoe. Even so, the past 30 days include only three major outages, compared with a typical five, suggesting the recent spike is notable but not unprecedented over a longer period.

Previously, we outlined the main factors behind Microsoft’s Ignite rollout of new AI models for coding and cybersecurity, including MAI Code 2 and a security-focused model.

This material may contain third-party opinions, none of the data and information on this webpage constitutes investment advice according to our Disclaimer. While we adhere to strict Editorial Integrity, this post may contain references to products from our partners.