Free Intensive Nov 21–22Free 2-day hands-on Vibecoding Intensive on Nov 21–22 at 16:00 MSK (Sat–Sun) · Live Cohort 3 starts Nov 23Join Free →Cohort 3 (Nov 23)
ReleaseInsight · On RadarTechCrunch→ original

Microsoft Services Down for Nearly 2 Days, FAA Halts Flights: Is AI to Blame?

Over two weeks, Apple Pay, NASA's internal communications network, and Microsoft Outlook, Teams and Copilot (for nearly two days) went down in the US, and the Federal Aviation Administration (FAA) halted all flights for several hours. Authorities have not confirmed a hack. Investor Dave Devault sees the cause in rushed patches following discoveries by AI scanners, while Anusha Ayer of Corsha points to hasty integration of new systems with outdated hardware. The takeaway: roll out patches and migrations in stages, with testing.

AI-processed from TechCrunch; edited by Hamidun News
Microsoft Services Down for Nearly 2 Days, FAA Halts Flights: Is AI to Blame?
Source: TechCrunch. Collage: Hamidun News.
Listen to article
Article Fact Sheet & Key Figures
ParameterDetails
Primary sourceTechCrunch ↗
Editor-in-Chief
MicrosoftOutlook, Teams and Copilot were unavailable for nearly two days
Key metric #2The FAA's full flight halt was only the third such event since the September 11 attacks
Key metric #3Devault expects a wave of startups focused on automated patching and safe updates

Over the past two weeks, one service after another failed in the US, services that everyday life depends on. Apple Pay stopped working. NASA's internal communications network went down. The Federal Aviation Administration (FAA) suspended all flights in the country for several hours, and this is only the third such case since the September 11 attacks. Microsoft's Outlook, Teams and Copilot assistant were unavailable for almost two days. The first thought is an AI-assisted hacking attack. But the official explanation is different, and the expert debate offers practical lessons for anyone rolling out updates or connecting new systems to old ones.

What is known about the outages

Recent background fuels the suspicion. A few weeks ago, hackers reached refrigerators in a grocery store and a blood bank in France. In addition, on October 21, during what was dubbed "Black October" for cybersecurity, a series of incidents affected PHP, Node.js, LiteLLM and other projects. Formally, those stories are not connected to each other.

The US Director of National Intelligence said the current outages do not look like state actions. President Trump called them "isolated events." Many companies themselves say the cause is technical, not a cyberattack. Investigations are still ongoing, so there are no final conclusions. TechCrunch only notes that so many outages in such a short period look too clustered to be coincidence.

Version one: AI finds holes too fast

Dave Devault is an investor who has worked in cybersecurity for more than thirty years and previously led McAfee. He believes the outages are most likely linked to AI, but not as a weapon used by attackers. In his words, AI tools such as Mythos uncover vulnerabilities at unprecedented speed and scale.

Imagine a scanner that overnight reveals hundreds of unlocked doors in a building that no one had noticed for years. The owner rushes to change the locks everywhere at once, before thieves exploit them. Some doors jam in the process, and the building's operations grind to a halt. According to Devault, that is what the current wave looks like: companies are massively patching long-standing holes, and this surge of patching is breaking working systems in the short term. He calls it the "growing pains" of a security system that is becoming stronger. The pain is real, he admits, but it is better for companies to find and close the holes themselves, even chaotically, than for attackers to do it.

Version two: not AI, but hasty stitching of old and new

Anusha Ayer, a programmer and founder of the security startup Corsha, disagrees. In her view, the rapid adoption of AI has not caused these outages so far. The culprit is a hasty mix of legacy and modern systems: industrial systems were never designed to be networked, and now they are being urgently connected to new ones without understanding the operational and cyber risks, and without the ability to protect the old part.

The FAA is an example. According to preliminary data, the agency modernized its telecommunications infrastructure, and the transition overloaded part of the outdated equipment. The new technology operated in the normal mode for modern systems, but the old hardware could not handle that load. It is like connecting a powerful new pump to pipes designed for lower pressure. Viewed this way, Ayer says, AI is being unfairly blamed.

Senator Mark Warner, vice chairman of the Senate Intelligence Committee, called the outages a warning sign: AI is developing faster than we manage to handle its consequences. At the same time, he stressed that the investigations are not finished.

What to do in practice

Both versions agree that the problem lies in the pace and order of changes. Several rules follow:

  • Do not roll out mass fixes all at once: determine what to fix first and what can wait.
  • Test patches before production. Devault says directly that deploying AI requires more testing.
  • Segment networks more strictly and restrict access to legacy systems, so that new connections are made in stages and risk is reduced (Ayer's advice).
  • Do not connect old and new systems until you have calculated how much load the old part can handle.

There is also a market side. Devault expects growth in the number of companies working on AI-based vulnerability management, autonomous remediation, safe testing and patch deployment, and the resilience of critical infrastructure systems. For founders, this is a notable niche: organizations need tools to move "at AI speed" without breaking things.

Limitations

These are expert opinions, not findings of an investigation. Officially, no direct link between the outages has been established, and the interviewees themselves disagree on AI's role. Moreover, both work in the security industry: Devault is an investor, and Ayer runs a startup in this field, so growing demand for such solutions benefits them. One thing is reliably known: over two weeks, several high-profile failures occurred, and managing change in complex systems has become harder.

Story

Key Questions Answered

What is known about the outages?

Recent background fuels the suspicion. A few weeks ago, hackers reached refrigerators in a grocery store and a blood bank in France. In addition, on October 21, during what was dubbed "Black October" for cybersecurity, a series of incidents affected PHP, Node.js, LiteLLM and other projects. Formally, those stories are not connected to each other. The US Director of National Intelligence said the current outages do not look like state actions.

What are the key details on Version one: AI finds holes too fast?

Dave Devault is an investor who has worked in cybersecurity for more than thirty years and previously led McAfee. He believes the outages are most likely linked to AI, but not as a weapon used by attackers. In his words, AI tools such as Mythos uncover vulnerabilities at unprecedented speed and scale. Imagine a scanner that overnight reveals hundreds of unlocked doors in a building that no one had noticed for years. The owner rushes to change the locks everywhere at once, before thieves exploit them.

What are the key details on Version two: not AI, but hasty stitching of old and new?

Anusha Ayer, a programmer and founder of the security startup Corsha, disagrees. In her view, the rapid adoption of AI has not caused these outages so far. The culprit is a hasty mix of legacy and modern systems: industrial systems were never designed to be networked, and now they are being urgently connected to new ones without understanding the operational and cyber risks, and without the ability to protect the old part. The FAA is an example.

ZK
Zhemal Khamidun — CPO at Alpina Digital, product manager of AlpinaGPT. Building AI products since 2019. Hamidun News is a pet project that grew into a production aggregator with 6,000+ readers.
✦ Hands-On AI Engineering · Hamidun Academy

Build & Ship Production AI Agents Without Boilerplate Code

Start your free 72-hour trial at Hamidun Academy (academy.hamidun.com) — interactive lessons on Claude Code & Cursor, live Tech Radar, and production agent templates.

Start 72h Free Trial →Vibe-Coding Intensive
What do you think?
Loading comments…