Apple and a hacker's future
- Source
- stratechery.com
- Author
- maguay
- Date
Context
A Stratechery essay uses a personal security incident to argue that Apple's platform fits badly with machines built to run AI agents. According to the essay, the author's always-on Mac Mini, used only to run AI coding agents, was hacked through a macOS screen-sharing flaw that Apple had already patched, after the machine's screen-sharing port was left reachable from the internet. The author takes the blame: screen sharing was on and the update was not installed. A coding running on the machine flagged the intrusion, and the author then wiped the machine.
The argument is that macOS permission prompts are built for a person at a screen. They can only be answered through the graphical interface, so an agent fails silently until someone logs in to click OK, which here meant keeping screen sharing enabled, a step the author says contributed to the breach. Apple has said it will add controls for granting Full Disk Access, which the author finds unsettling. The author says they bought a Linux server and will not put another Mac in a rack, but are not predicting Apple's downfall. This is one commentator's opinion, and the account of the incident is the author's own.
- AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
Checking sign-in…
Loading comments…
