What a week. Day after day, security pros had to cope with a seemingly endless onslaught of unconnected outages.
It was hard to keep up and stay poised and frankly, not freak out!
First Microsoft 365 went down in more than 70 countries, then the airlines ground to a halt. Those outages were followed by major banks like JPMorganChase as well as outage at NSANet, the National Security Agency’s classified internal computer network.
While everyone has been struggling to make sense of it, most security pros agree that we’ve arrived at a “new normal” for the AI era: A time in which vulnerabilities are being discovered faster than teams can patch, and because of the connected nature of these systems, a single failure can cause widespread outages.
More of these outages are expected before the industry reaches an equilibrium.
“While it’s unusual to see this many outages in the same week, the bigger story is how much disruption a relatively small number of failures can cause,” said Matt Hartman, chief strategy officer at the Merlin Group. “We’re also layering AI onto digital infrastructure that was already showing signs of strain. AI isn’t necessarily causing these failures, but the speed and scale of its adoption could amplify weaknesses we’ve spent decades accumulating.”
Hartman also noted that there’s no evidence right now that these incidents are connected or part of a coordinated cyberattack — and we should be careful not to assume they are. Establishing that it was a coordinated cyberattack would require common attack infrastructure, similar malicious activity, and forensic evidence tying the disruptions together.
“Even without that, what we do know is concerning enough: our dependence on a small number of technology providers means a single failure can disrupt essential services without a cyberattack ever taking place,” said Hartman.
Bri Frost, director of product management at Cloud Range, said from Microsoft and air travel to banking, people are seeing major disruptions unfold day after day. It’s understandable that they’re asking whether these events are connected and whether the systems they rely on are secure.
“When critical systems go down, the first public statement almost always says it wasn’t a cyberattack,” said Frost. “These statements usually come out before the investigation is done. They may turn out to be right, but announcing a conclusion that early costs trust, especially when several major disruptions land in the same week. People notice the pattern, and ‘not cyber’ with no further detail doesn’t answer their real question. That question: whether the systems we depend on can hold up and whether the people running them are ready when they don’t. Organizations would build more trust by saying what they know, what they’re still investigating, and when they’ll share more, then follow through.”
The outages underscore a general weakness
Justin Beals, founder and CEO of Strike Graph, said he believed that two years of deep cost cutting have thinned the teams that support critical infrastructure.
“No operator has said that, and I can’t prove it from the outside,” said Beals. “What the record shows is small changes cascading. Microsoft’s preliminary report traces its outage to a defect in a change deployed in July, and says its automated recovery was not designed for a failure that broad.”
Transportation Secretary Sean Duffy said the FAA ground stop began with a router being replaced: what’s unusual is how small the triggers were compared with the damage.
“Verizon’s 2026 Data Breach Investigations Report found the median time to fully patch rose to 43 days in 2025, from 32 the year before,” said Beals. “So, AI is shortening the time from disclosure to a working exploit, so that gap pushes teams to cut testing and move changes into production sooner. Microsoft’s preliminary report shows what testing can miss. It says the defect passed load testing because the lab hardware was weaker than production. The outage risk is highest where rollback is hardest: core network equipment, legacy systems and isolated classified networks that cannot be staged the way a cloud service can.”
Aharon Twizer, co-founder and CEO of ControlMonkey, added that we need to accept that outages will happen, even with the biggest cloud and technology providers.
The question: How prepared are we when they do?
“Companies invest heavily in preventing downtime, but often don’t invest enough in recovery,” said Twizer. “Having backups is important, but backups alone don’t mean we can bring the business back online. We also need the network, identity, security, and infrastructure configurations that make everything work. Organizations should test their recovery plans against real scenarios, not just assume they’ll work.”
Twizer said companies need to check whether their disaster recovery plan actually covers the infrastructure configuration, not just the data.
Here are the questions to ask: If the company’s DNS records disappear, its identity permissions are accidentally changed, or the networking configuration breaks, can the team restore it? Does the team know what the last working configuration looked like?
After that, Twizer recommended testing the recovery process, including scenarios in which a cloud region or third-party service isn’t available.
“And with AI increasingly making changes to infrastructure, we need to automate recovery as well,” said Twizer. “We can’t have infrastructure changing automatically while recovery still depends on engineers manually rebuilding everything during an incident.”
Conspiracy theories abound as people feel vulnerable
Chris Hughes, vice president of security strategy at Noma Secuirty, added that the timing of these outages is unusual because we're seeing concurrent outages and incidents across both the public and private sector, which understandably creates public concern and a lot of speculation, including conspiracy theories.
“What it actually highlights is the digital fragility of the ecosystem we all depend on,” said Hughes. “Everything from consumer goods to critical infrastructure to national security systems runs on software, and that software has accumulated decades of security technical debt. AI has industrialized vulnerability discovery, organizations are racing to fix what's being found, and the patching itself can disrupt operations. That's the availability piece of the CIA triad.”
We've seen industrialized vulnerability discovery all year, in programs like Patch the Planet and Project Glasswing and in a steady stream of findings from frontier labs and researchers. Vulnerability trackers reported the 2026 CVE count passing 100,000 earlier this month, more than double last year and a record. That volume has outpaced the ability to patch and test safely, and when teams compress testing windows to keep up, they raise the risk of exactly these kinds of outages.
“The work that matters is risk-based prioritization around exploitability, known exploited vulnerabilities, and reachability, along with better testing, rollback plans, and resilience,” said Hughes.
The coincidences are hard to ignore, and that’s why the speculation has taken off, but Hughes said the better-supported story right now is accumulated technical debt and rushed remediation under the pressure of industrialized discovery: organizations should still prepare for agentic offensive frameworks, even if this cluster isn't ultimately attributed to them. Security teams need to assume the discovery-and-exploitation cycle now runs at machine speed.
“That still comes down to doing the basics exceptionally well: hygiene, posture, and vulnerability remediation, with risk-based prioritization rather than trying to patch everything at once,” said Hughes. “Inventory internet-facing systems, including forgotten partner portals and legacy services, shorten the life of standing credentials, and put real testing and rollback behind accelerated patching so availability doesn't become the casualty. On the response side, monitor for behavior and velocity rather than static indicators, and invest in recovery and resilience, because some weaknesses will still be there when an incident starts.”
Louis Eichenbaum, Federal CTO at ColorTokens, added that the most concerning aspect of these outages isn't necessarily their timing: it's what they reveal about the fragility of our increasingly interconnected digital infrastructure. Government agencies, financial institutions, airlines, and technology providers may operate independently, but Eichenbaum said many rely on the same underlying cloud platforms, identity services, telecommunications networks, and software supply chains.
Eichenbaum noted that we've spent decades improving efficiency through consolidation and interconnected systems, but that efficiency has also introduced concentration risk: a single failure in a shared dependency can cascade across multiple sectors, creating disruptions that resemble a coordinated cyberattack even when no malicious activity was involved.
The fundamental question: Whether we've prioritized efficiency and connectivity at the expense of resilience?
“Critical infrastructure must be designed to withstand failures, not simply prevent them,” said Eichenbaum.
The focus turns to availability and resilience
Casey Marks, chief operating officer at ISC2, explained that the CIA triad of confidentiality, integrity and availability recognizes that protecting information and keeping critical systems accessible are both essential to cybersecurity.
“The recent outages have put availability to the test and give organizations a reason to take a hard look at their business continuity and recovery plans,” said Marks.
Marks said teams should start by identifying the systems the organization can’t operate without and the vendors those systems depend on. Then test what happens when those services go down. Teams should know who can make recovery decisions, how to reach employees and customers if their usual channels are down, which operations can continue, and which systems need to come back first.
“A technical failure can stop a business just as effectively as an attack, even when no data has been stolen,” said Marks. “Planning for disruption and testing how the company will recover deserve the same attention as preventing attacks.”
John Harbaugh, CISO and CTO at BlueVoyant, added that with the growth and adoption of AI-assisted maintenance, upgrades, and software development, the industry needs to turn resilience into something it can demonstrate: organizations should understand which critical services depend on the same providers, test what happens when those dependencies fail and establish clear authority to stop a deployment or roll back a change. A recovery plan needs to work when a major provider is unavailable. As AI takes on more software development and operational work, those same controls need to apply to AI-assisted changes.
Moving forward, Harbaugh said security teams need humans and AI agents with clearly defined roles, pre-approved actions, and escalation play books. We need to support that structure via evidence that lets teams independently verify what an agent actually did, and offer critical visibility into what AI agents can access and change. This requires people trained to evaluate these decisions, with the ability test the agents in ways that can contain or reverse their actions.
“This is not the time to be afraid of AI,” said Harbaugh. “These are very complicated and unique architectures and partnerships with very diverse technical stacks and touchpoints. AI, with the right controls and right human expertise integration, is truly necessary to enable the right speed-to-action in these critical networks.”
Heath Renfrow, co-founder and CISO at Fenix24, explained that we are entering a period in which AI has accelerated both sides of the equation: attackers can discover and exploit weaknesses faster, while defenders are being pushed to respond at unprecedented speeds.
Historically, organizations had more time to evaluate vulnerabilities, test patches, understand dependencies, and deploy changes in a controlled manner, but Renfrow said AI has compressed that entire timeline.
Today, security teams are under tremendous pressure to remediate vulnerabilities before attackers can exploit them, and that urgency can come at the expense of adequate testing and operational validation. Renfrow said the question is not simply whether the company can protect its infrastructure. The question is whether the organization can continue operating when those systems become unavailable.
“Resilience is not exclusively about surviving a ransomware attack or outage,” said Renfrow. “It’s about surviving the loss of technology the business depends on, regardless of whether the cause is a cyberattack, human error, infrastructure failure, or a third-party outage. Your business does not care why its technology is unavailable. It cares how quickly it can operate again.”