Security leaders split on whether AI patching caused US outages

After outages at Microsoft 365, the FAA, JPMorgan Chase and the NSA, industry figures debate how far AI-accelerated vulnerability discovery and rushed fixes are to blame.

SecurityPublished
What happened
SecurityWeek collected industry reactions to recent US outages and to claims that AI-driven patching pressure contributed to them.
What’s new
Experts disagree on AI's role, but most call for tested rollback plans rather than slower patching.
Why it matters
Organisations face more fixes than ever and must weigh the risk of leaving flaws open against the risk of breaking live systems.

Several major US outages recently hit email, air travel, banking and government systems. They included a Microsoft 365 outage lasting nearly a day, a nationwide FAA ground stop, a JPMorgan Chase problem that slowed transactions at other banks, and a failure on the NSA's classified internal network. According to SecurityWeek, officials reported finding no sign of a cyberattack and no connection between the incidents. Microsoft blamed a software defect several months old, the FAA pointed to a router it had just installed, and the NSA attributed its problem to routine maintenance work.

Jay Clayton, the Director of National Intelligence, gave a different account, tying the NSA disruption to patching for a PHP flaw found with AI assistance. Sen. Mark Warner said AI-driven discovery is forcing organisations to patch many systems simultaneously.

Industry figures quoted by SecurityWeek are divided on how much this explains. Azul CTO Gil Tene said there is strong reason to believe patching at AI-accelerated speed is already behind major incidents. He noted that the Java community absorbed four update events in four weeks, and he argued for security-only updates. Catalogic CEO Ken Barth cited more than 1,400 vulnerabilities fixed in Microsoft's October release and 1,094 in November, but said nothing published links that volume to the Microsoft 365 outage. OPSWAT's Benny Czarny said nothing can be attributed to AI without evidence from investigators.

Most commentators converged on resilience: staged changes, verified restore points and rollback plans that have actually been tested. Huntress adviser Bryson Byrd warned that organisations may start holding back patches "out of fear of being the next Microsoft or FAA."