Theori's 'Xint Code' Discovers CVSS 9.8 Vulnerability in Apple OS
Published2026.08.05 08:46
'Xint Code' Reports 2 Cases Including CVSS 9.8 Kernel Vulnerability
Theori, an offensive cybersecurity specialist company, announced that it has discovered security vulnerabilities in Apple's operating systems through its AI-based source code security analysis solution and has led to the application of official security patches.
On the 5th, Theori announced that two vulnerabilities discovered by its AI source code security analysis solution 'Xint Code' in Apple's major operating systems have been reflected in Apple's recently published security updates.
The most noteworthy vulnerability addressed this time is the kernel vulnerability 'CVE-2026-64775,' which is evaluated as Critical with a CVSS (Common Vulnerability Scoring System) score of 9.8.
According to Theori, this vulnerability is a flaw that allows a specific application to abnormally terminate the system without requiring separate user interaction. Apple explained that it has fixed problems occurring in the memory initialization process and improved memory handling methods to block related risks.
The concurrently fixed 'CVE-2026-64744' is a vulnerability at the CVSS 5.5 level, known as an issue where a specific application can expose kernel memory information. Apple resolved this by applying additional validation procedures.
This security update has been applied across major Apple platforms including △iOS 26.6 △iPadOS 26.6 △macOS Tahoe 26.6 △macOS Sequoia 15.7.8 △macOS Sonoma 14.8.8 △tvOS 26.6 △visionOS 26.6 △watchOS 26.6.
Theori evaluated this Apple vulnerability discovery as another case demonstrating the scope of application for AI-based source code vulnerability detection technology.
The company previously disclosed that it had discovered the Linux kernel privilege escalation vulnerability 'Copy Fail (CVE-2026-31431)' and three high-risk Android vulnerabilities (CVE-2026-0095, CVE-2026-0138, CVE-2026-0143).
'Xint,' developed by Theori, is an AI-based security platform providing source code analysis, penetration testing, and security validation capabilities.
Among these, 'Xint Code' is introduced as a white-box security analysis solution that detects potential vulnerabilities by analyzing the structure and data flow of source code.
Kwak Kyung-joo, Head of Theori's Xint Product Division, stated, "This case demonstrates that the AI engine can discover security vulnerabilities that are difficult for humans to find even in large-scale source code environments."
Related news
Theori Participated Research Adopts NDSS… Presents New Path for Linux Kernel Attacks
A Linux kernel attack technique research involving cybersecurity firm Theori has been accepted by the international security conference NDSS. The core of this research is presenting new attack possibilities through data manipulation methods in a situation where existing control flow-based attacks have become difficult due to the recent introduction of strong defense mechanisms in the Linux kernel. The research team simplified the attack procedure through the 'DirtyFree' technique, which injects fake objects by exploiting forced memory deallocation, and confirmed applicability to multiple actual kernel vulnerabilities. Simultaneously, the team proposed a defense mechanism that minimizes performance degradation, demonstrating the direction of kernel security research.
2026.03.18by 명세환 기자
Theori Unveils AI-Based Static Analysis Tool 'Zint Code' at RSAC 2026
Theori will unveil "Zint Code," an AI-based static application security testing tool, at RSAC 2026 held in San Francisco, USA. The product is characterized by its ability to analyze large-scale source code, configuration files, and binaries in a short timeframe to identify vulnerabilities. Unlike conventional rule-based static analysis tools, it is designed to understand code context and business logic, thereby reducing false positives and presenting both actual exploitability and impact scope. Through integration with its existing dynamic analysis solution "Zint Web," Theori presents a comprehensive security framework that encompasses everything from external attack detection to code-level root cause analysis.
2026.03.23by 배종인 기자
Teori Selected as AI Voucher Supplier… Supplying LLM Security Solutions
Cybersecurity company Theori has been selected as a supplier for the 2026 AI Voucher Support Program. The company plans to address information leakage and regulatory compliance issues that arise during generative AI adoption through its LLM security solution 'AlphaPrism'. The solution is designed to analyze the context of prompts and source code to identify sensitive information such as personal data and authentication keys, and to continue AI utilization through masking methods rather than simple blocking. The government's AI Voucher Program supports demand enterprises with up to 200 million won per project for solution purchase costs. Based on recent application cases at POSCO DX, Theori plans to respond to expanding demand for enterprise AI security.
2026.03.24by 명세환 기자
Theori and Aton Sign Agreement at RSAC 2026… Pursuing Global Expansion of AI Code Security Solutions
Offensive cybersecurity firm Theori and fintech security company Atonomy have signed a business agreement at RSAC 2026 held in San Francisco, USA, and are moving forward with expanding their enterprise security business. The two companies plan to jointly pursue global marketing and market development centered on Theori's AI-based Static Application Security Testing (SAST) solution 'Zint Code'. This collaboration can be viewed as a case of combining technological capabilities and business networks to penetrate overseas markets amid growing demand for code security solutions utilizing AI.
2026.03.30by 명세환 기자
Teori CEO Park Se-jun Receives Commendation from Minister of the Interior and Safety
TiorI's CEO Park Se-jun, an offensive cybersecurity company, received a commendation from the Minister of the Interior and Safety for his contributions to strengthening the nation's information protection capabilities at the opening ceremony of the 32nd Information and Communications Network Information Security Conference (NetSec-KR 2026).
2026.04.16by 배종인 기자
