Anthropic recently announced Project Glasswing, an initiative that allows tech companies like Apple to use its new Claude Mythos Preview frontier AI model to detect security vulnerabilities in operating systems and web browsers.
The Wall Street Journal reported today that researchers at cybersecurity firm Calif used Claude Mythos Preview to discover a new macOS security vulnerability last month. Specifically, they used the pattern to write code connecting two macOS bugs in a way that resulted in what’s known as an elevation of privilege exploit.
Security researchers said the exploit would not have been possible with Mythos alone because it still required their human expertise, but it nonetheless proves that AI can help uncover software vulnerabilities.
Apple said it was reviewing Calif’s report to validate the findings.
“Security is our top priority and we take reports of potential vulnerabilities very seriously,” an Apple spokesperson said. The Wall Street Journal.
It’s unclear if Apple has yet patched the exploit. Apple’s security notes for the macOS 26.5 update released this week mention a fix for a kernel-level vulnerability, and they credit Calif and Anthropic for discovering it. Yet the report says California only met with Apple this week and suggested a fix was still coming.
We have contacted Apple for comment.
Popular stories
Apple warns that Canada’s Bill C-22 could impose encryption backdoors
Apple and Meta have opposed a Canadian bill that the companies say could force them to create backdoor access to encrypted user data if passed by the country’s parliament. Proposed by Canada’s ruling Liberal Party, Bill C-22 contains provisions that could be similar to a British data access order sent to Apple last year, depending on how they are implemented. Back in February…
Anthropic rebuilds Claude Code desktop app around parallel sessions
Anthropic has released a redesigned Claude Code experience for its Claude desktop app, bringing a new sidebar for managing multiple sessions, a drag-and-drop layout for organizing the workspace, and more. The new sidebar displays every active and recent session in one place, and users can filter by status, project, or environment, with the ability to group sessions by project. A new side….
Anthropic launches Claude Design for the creation of prototypes, pitch decks and models
Anthropic today launched Claude Design, a new AI product for creating designs, prototypes, slides and more. Claude Design uses Opus 4.7, a new AI model introduced earlier this week. Opus 4.7 is Anthropic’s most capable vision model, and it can see images with higher resolution. Anthropic says he is “most tasteful and creative” when performing professional tasks. He is able to create…