
WEBSITES AFFECTED
70%
DAYS OF OUTAGES
5
MAJOR SYSTEMS DOWN
4+
INNING · PER MANDIA
1st
Over the past five days, Microsoft 365, JPMorgan Chase, and Charles Schwab all suffered significant service outages. The FAA issued a national ground stop affecting every US airport. The natural instinct is to blame coordinated cyberattacks or runaway AI. The actual cause is more mundane and more structurally interesting: organizations are patching software faster than their systems can absorb the changes.
Extraction Map
AI tools have gotten very good at discovering security vulnerabilities — flaws that human-written code has harbored for years, sometimes decades. Last month, AI identified a massive vulnerability in a programming language used by up to 70% of websites. The discovery triggered what Senator Mark Warner called a "mad scramble" to patch, and what Armadin CEO Kevin Mandia described as update speeds that outrun testing capacity. The patches themselves are destabilizing the systems they're meant to protect. The pattern is a classic feedback loop. AI finds vulnerabilities at machine speed. Organizations push patches at machine speed. But the downstream systems receiving those patches — legacy infrastructure, interconnected services, production environments — still operate on human timescales. The mismatch creates cascading failures. As former cybersecurity reporter and VC founder Nicole Perlroth noted, rapid-fire fixes are more likely to cause failures "in unexpected, cascading ways." Senator Warner, vice chair of the Senate Intelligence Committee, confirmed these outages are not the product of coordinated nation-state cyberattacks. Transportation Secretary Sean Duffy attributed the FAA ground stop specifically to aging air traffic control infrastructure. The reassurances are real but beside the point: the systemic risk isn't adversarial, it's operational. Organizations are creating fragility in the name of security. Mandia frames this as "the first inning" of an AI-era adjustment. The long-term thesis is straightforward: AI-discovered vulnerabilities get patched, systems become more secure, and organizations develop operational maturity to handle the pace. The short-term reality is that enterprises are running a live experiment on their production systems, and customers are bearing the cost of every failed patch cycle. The broader concern is what happens between now and that mature state. A crypto heist, refrigerator hacking exploits, and an attack on French blood plasma refrigeration systems have all followed in the wake of the initial vulnerability disclosures. The window between AI-discovery and successful patching is exactly when systems are most exposed — both to the original vulnerability and to the patch-induced instability. Adversaries know this timeline too. The uncomfortable truth is that every organization now faces a lose-lose choice on tempo: patch slowly and remain vulnerable to AI-speed exploits, or patch fast and risk self-inflicted outages. The enterprises that solve this sequencing problem first will have a genuine competitive advantage. Everyone else will keep crashing.
Worth reading?
SKIP
Score
36
/100
Extractive in the short term — consumers and businesses absorb outage costs while cybersecurity firms and AI vendors capture the upside of faster vulnerability detection without bearing downtime consequences.
What Is GRIN?
GRIN stands for Generativity, Resilience, Innovation efficiency, and Novelty openness. These four dimensions capture the essential characteristics of any memetic intelligence — whether a nation, a corporation, a religion, or a movement.
About GRINGenerativity
2/5
Resilience
1/5
Efficiency
1/5
Novelty
3/5
Extractive
3/5
Inverted
Does this create new capability or just redistribute existing value?
Long-term generative potential is real: AI finds bugs humans missed for years. But today's value creation is negative — patching at speed destroys uptime faster than it creates security.
Verified Facts (7)
Kevin Mandia, CEO of cybersecurity company Armadin, stated on CNBC's Squawk Box that updates are happening too fast to test all outcomes
Senator Mark Warner, vice chair of the Senate Intelligence Committee, confirmed outages are not from coordinated cyberattacks
Transportation Secretary Sean Duffy attributed FAA ground stop to aging air traffic control infrastructure
AI discovered a programming language vulnerability affecting up to 70% of websites last month
Microsoft 365, JPMorgan Chase, and Charles Schwab all experienced outages within a five-day period
Nicole Perlroth, former cybersecurity reporter and VC fund founder, wrote on X that rapid patches are more likely to cause failures in unexpected, cascading ways
A crypto heist at Binance, refrigerator hacking, and an attack on French blood plasma refrigeration (EFS) followed the initial vulnerability disclosures
More latest news
Read all