Microsoft Plugs Nearly 400 Security Holes
Microsoft has released updates to address 398 security vulnerabilities, including one actively exploited and two previously disclosed. The increase in patch frequency is attributed to AI-assisted vulnerability discovery, though patching remains a human-centric process.
Microsoft Addresses 398 Security Vulnerabilities in August Patch Update
Microsoft has released a significant update addressing 398 security vulnerabilities across its Windows operating systems and supported software. Among these, one vulnerability is already being actively exploited, and two others were disclosed publicly before the patch was released. This update continues a trend of increasing patch frequency, driven in part by the use of artificial intelligence in vulnerability discovery.
The August update does not break the record set in July, when Microsoft released over 570 security updates, but it does double the number of patches from June. The company has attributed the surge in vulnerability discoveries to AI tools that are now being used by security researchers and organizations to identify flaws more efficiently.
Of the 398 vulnerabilities addressed, 42 were rated as “critical” by Microsoft, indicating that they could allow attackers to gain remote control of a system with minimal user interaction. One of the most notable fixes is CVE-2026-68820, a privilege escalation flaw in the afd.sys driver, which Automox describes as a critical component in Windows socket connections. This vulnerability, while requiring a specific sequence of events to exploit, is already being used in attacks.
Another critical flaw, CVE-2026-62832, is in the Windows User Profile Service and may be related to the “LegacyHive” disclosure from the bug hunter known as Nightmare Eclipse. Microsoft has also addressed CVE-2026-72971, a low-impact local tampering vulnerability that is unlikely to be exploited.
While AI is proving effective in identifying vulnerabilities, the process of patching remains largely dependent on human oversight. Researchers at 1Password have found that AI-generated patches often fail to resolve the issue or introduce new vulnerabilities. Experts like Ed Skoudis of the SANS Technology Institute emphasize the need for human involvement in testing and refining AI-generated fixes.
Security professionals are advised to approach the increased patching cadence with caution. Tyler Reguly of Fortra notes that while the volume of patches is rising, only one of the 398 vulnerabilities is currently known to be actively exploited. Organizations are encouraged to evaluate their patching workflows and ensure that updates are thoroughly tested before deployment.
As AI continues to play a larger role in both identifying and addressing vulnerabilities, the balance between automation and human oversight remains critical. While AI can significantly enhance the speed and scope of vulnerability detection, the reliability of AI-generated patches is still under scrutiny. Security teams must continue to prioritize testing, iteration, and verification to ensure that patches are both effective and safe.
Source: Krebs on Security. Analysis based on publicly available information; no hands-on testing is claimed.
