A viral post on X recently claimed that an OpenAI model identified 35 vulnerabilities in Apple devices, potentially allowing remote control. However, official Apple security documentation does not support this figure. Apple’s records confirm that the company’s Codex Security tool was credited for only three specific vulnerabilities within the WebKit engine.

The confusion likely stems from Apple’s June 2026 security updates for iOS, iPadOS, macOS, and Safari, which addressed over 30 flaws using AI-assisted tools. Apple accelerated its patch release schedule in response to the growing risk of AI-driven hacking techniques, marking a shift in how the company manages its security pipeline.

Separately, in November 2026, OpenAI reported that its autonomous cybersecurity pipeline identified 948 distinct local privilege-escalation vulnerabilities in the Linux kernel. This operation, conducted in October, reportedly cost $1.2 million in API credits and included the generation of both exploits and patches for the discovered flaws.

While the viral claim of 35 bugs remains unverified, the events highlight the increasing role of automated tools in identifying software vulnerabilities. Both Apple and OpenAI are integrating these capabilities into their standard security workflows, as the window between the discovery of a flaw and its potential exploitation continues to shrink.