Thousands of North American schools and universities were paralyzed Thursday after a ransomware attack on a key educational software provider. The hackers — called ShinyHunters, believed to be mostly Western teenagers behind several other high-profile attacks — demanded crypto payments to unlock the system, and threatened to leak private data. The attack is likely AI-assisted, and is a harbinger of cybersecurity’s chaotic future.

Claude’s Mythos AI model was recently able to detect hundreds of previously unknown vulnerabilities in browsers and websites; Mozilla has confirmed that those found in its Firefox browser were almost all real. Mythos is not publicly available, but comparably cyber-capable Chinese open-source models will be soon, and hackers will be able to find any holes not plugged by then.