Security firm Calif.io demonstrated an untethered jailbreak on an iPhone 18 Pro running iOS 27.0.1 at the Hexacon conference in Paris on October 16, 2026. This marks the first successful permanent jailbreak on such devices, allowing root access to persist after a reboot without requiring a connected computer. The demonstration, captured in video by researcher Billy Ellis, showed the team unlocking the phone with a standard passcode and opening an SSH session as root, confirming full system control.
Calif.io has not disclosed the specific vulnerabilities used or released a public tool, stating that details will be published "in due time." The firm’s newsletter also listed other demos, including exploits for Signal and iMessage. Apple’s iOS 27 security notes from September 14 credit Calif.io, working with Claude and Anthropic Research, for two unrelated crash bugs, but do not link these to the jailbreak chain. Apple has not commented on the demo, maintaining its policy of not confirming security issues until patches are available.
The achievement highlights the continued vulnerability of Apple's locked-down ecosystem despite heavy investment in signed code, app sandboxes, and protected startup chains. While few users currently utilize jailbreaking, the existence of a working exploit on current hardware demonstrates that defenses like Pointer Authentication Codes can still be bypassed. The exposure of iOS 27.0.1 remains dependent on when Calif chooses to release technical details, as Apple typically patches such chains once informed.
The demonstration of an untethered jailbreak on the iPhone 18 Pro signifies a critical breach in Apple's most recent security architecture. Unlike tethered methods that require external tools upon restart, this persistent root access indicates a sophisticated exploit chain capable of circumventing hardware-backed protections and secure boot validations. This capability is particularly concerning because it mirrors the techniques often employed by commercial spyware vendors targeting government officials and high-profile individuals, suggesting that the barriers to deep-level device compromise remain surmountable even on cutting-edge hardware.
A notable aspect of this development is the explicit collaboration between human researchers and AI models, specifically Anthropic’s Claude. Previous statements from Calif indicate that AI assistance accelerated the identification of bugs and exploit development, reducing the timeline for complex tasks like building macOS kernel exploits. This integration of generative AI into offensive security workflows may lower the barrier to entry for discovering zero-day vulnerabilities, potentially increasing the frequency and speed of such discoveries. As Apple awaits the publication of details to issue patches, the window of exposure for iOS 27.0.1 users remains open, underscoring the ongoing tension between closed-system security models and advanced exploitation techniques.
