Cyber Security News

Hacking Editorial Brief — August 12, 2026

AI-Driven Hacking Tools and Autonomous Attack Capabilities

Security researchers have demonstrated that open-source AI agents can be configured to function as coordinated hacking teams capable of running simultaneous reconnaissance and exploitation activities. The development has intensified concerns about autonomous AI offensive capabilities, prompting House Democrats to demand hearings and answers from major AI developers about incidents involving self-directed hacking by AI models. Rapid7 disclosed that AI assistance was used in discovering and chaining two SharePoint vulnerabilities (CVE-2026-55040 and CVE-2026-63520) to achieve unauthenticated remote code execution, marking a notable case of AI-augmented vulnerability research leading to critical exploit chains.

Active Zero-Days and Microsoft's Largest Patch Tuesday

Microsoft's August 2026 Patch Tuesday addressed 400 vulnerabilities, including CVE-2026-68820, a Windows Ancillary Function Driver for WinSock zero-day actively exploited in the wild. Zoom patched CVE-2026-53413, a critical zero-click remote code execution flaw in its annotation feature that allows meeting participants to execute code on other participants' devices without interaction. Metabase released urgent patches for a SQL injection vulnerability exploited as a zero-day, enabling unauthenticated remote attackers to obtain administrative access. Additionally, U.S. and South Korean agencies issued a joint warning about Gunra ransomware-as-a-service, which has been recruiting ethical hackers and deploying North Korean government-linked tools to target critical infrastructure. Iran-linked threat actors expanded attacks on U.S. water infrastructure, with confirmed intrusions now spanning at least 12 states including newly reported incidents in New Jersey and Alabama.

Sources: Financial Times · CNBC · The Hacker News · Bleeping Computer · Security Week · Security Affairs · FDD