News
cybersecurity Critical PHP Security Flaws Prompt Warning Ahead of November 19 Patches - CyberPress threat-intel Chinese APT Group Uses Silent EggStreme Backdoor to Maintain Hidden Access to Government Networks - GBHackers threat-intel WordPress 7.1.5 Patches Critical Privilege Escalation, SQL Injection and Cross-Site Scripting Vulnerabilities - GBHackers threat-intel Microsoft Introduces Hardened Server Core in Windows Server 2028 With Hundreds of Security Settings Enabled by Default - GBHackers cybersecurity AuditTeam Ransomware Gang Poses as Security Auditors and Use AV Drivers to Kill EDR Process - Cyber Security News cybersecurity Windows Hello Vulnerability Lets Attackers Brute-Force PINs in Just 40 Hours - CyberPress cybersecurity Hackers Actively Exploit Critical Apache Roller RCE Flaw to Execute Code Without Authentication - Cyber Security News threat-intel Windows Hello PIN Security Bypassed Using TPM Vulnerability, Researchers Crack 4-Digit PIN in 40 Hours - GBHackers threat-intel Hackers Use OpsAgent Malware to Disable Antivirus Protection and Launch Ransomware Attacks - GBHackers cybersecurity Microsoft Introduces Hardened Windows Server 2028 With Hundreds of Security Protections Enabled by Default - CyberPress threat-intel NGINX2Root: Five Critical Vulnerabilities Allow Unauthenticated Attackers to Execute Commands on NGINX Servers - GBHackers cybersecurity WordPress 7.1.5 Fixes 20 Security Flaws, Including Unauthenticated RCE and Privilege Escalation - CyberPress threat-intel PHP Warns of Critical Security Vulnerabilities, Urges Users to Prepare for Emergency Updates - GBHackers cybersecurity PHP Warns of Critical Security Flaws Affecting All Supported Versions, Fixes Coming November 19 - The Hacker News cybersecurity Hackers Abuse CyberPower UPS Software to Execute SYSTEM Commands and Dump Active Directory Data - Cyber Security News
1 / 15
All news ›
Cybersecurity

This is how AI is changing vulnerability discovery

Security researcher Florian Roth warns that new critical Linux vulnerability CVE-2026-98432, a flaw introduced in October 2015 by a performance optimization in the Linux TCP/IPv4 networking stack, shows how AI could change vulnerability discovery. He reports that Anthropic says its Claude model found the flaw and built a working exploit allowing remote code execution as root without authentication against a server with an MPTCP listener. Roth argues that AI makes such discovery easier to scale, so long-standing code in widely used software may face new risk.

Continue on x.com ↗ ‹ All news Get the newsletter

The summary and image come from Pro CISO; the image is shown from their website.