Security researchers have given the Fruity Cargo Cult Apple’s security halo a good kicking, with AI helping expose MacOS flaws.
According to the Wall Street Journal, researchers at Palo Alto security outfit Calif found two bugs in Job’s Mob’s desktop operating system while testing an early version of Anthropic’s Mythos AI software in April.
The discovery will annoy the tame Apple press, which has spent years pretending Job’s Mob kit is protected by fairy dust rather than engineering.
Calif’s researchers said their software linked two bugs with a handful of techniques to corrupt Mac memory. That let them gain access to parts of the device that should have been off limits, which is not quite the polished security miracle Cupertino likes to flog.
The attack is known as a privilege escalation exploit. If chained with other attacks, it could allow a hacker to seize control of the computer and make a mockery of the usual “it just works” chanting.
Former Google security researcher Michał Zalewski, who reviewed the Calif research but was not involved in the testing, said the technique mattered because Job’s Mob had spent so much effort locking down MacOS.
Job’s Mob, which is deploying and testing frontier AI models to test and patch vulnerabilities, is reviewing the Calif report to validate its findings.
“Security is our top priority, and we take reports of potential vulnerabilities very seriously,” a company spokeswoman said.
That line is wheeled out every time the magic orchard discovers that security through smugness is not quite a product category. The bug-finding capabilities of recent AI models from Anthropic and OpenAI have improved sharply in recent months.
Cybersecurity experts are now warning of a “Bugmageddon”. Earlier this year, Anthropic’s AI found more than 100 high-severity vulnerabilities in the Firefox browser during a two-week period.
That is about as many as the rest of the world usually finds in two months. In September 2025, Job’s Mob said it had turned its hardware and operating system expertise into a technology called Memory Integrity Enforcement, or MIE.
It described MIE as the culmination of an unprecedented design and engineering effort, spanning half a decade. Five days with Claude was enough for Calif to build code exploiting the two MacOS bugs, which rather takes the shine off the half-decade victory lap.
Calif chief executive Thai Duong said the attack could not have been pulled off by Mythos alone.
He said it needed the human security skills of Calif’s hackers, because Mythos is better at reproducing attacks already documented by humans.
“We haven’t seen cases where it comes up with new attack techniques. This is kind of a new thing,” Duong said.
