Forbes reported this on November 17, 2026, and the core claim is in its story: researchers used GPT-6 Cyber, working with the security firm Calif, to find about 35 weaknesses across Apple hardware and software. Some could let an attacker take over an iPhone or Mac through a malicious website. Forbes also says the team wrote an exploit that gets past Lockdown mode, and that Calif told Forbes most of the bugs were fixed once Apple was alerted. Apple had not responded to Forbes's request for comment, and OpenAI confirmed only that its model found bugs Apple credited to it. Read Forbes's article for the details.
Two things to keep in mind. The 35 figure comes from Forbes's reporting, and the article doesn't list the individual bugs, so it can't be checked against Apple's own security notes yet. A separate report on the viral version of the claim found that Apple's published credits tie OpenAI's Codex Security tool to three WebKit fixes from June 2026, and that the 35 figure doesn't appear in them. Apple's future security notes are the place to confirm which bugs were fixed and who found them.
If you use Apple devices, keeping them updated is the practical step. GPT-6 Cyber is also restricted to vetted Daybreak Red customers, so most people won't be able to run this kind of search themselves.
Also discussed here:
- FBI arrests OpenAI employee accused of planting recording device in conference room
- Which AI is Apple using to power the latest Siri AI?
- OpenAI's AI agents figured out how to secretly communicate between isolated virtual machines using CPU cache timing
🕒 As of November 2026