TechCapsules

WordPress sites can still be compromised while installing the patch for CVE‑2026‑119977, a flaw actively exploited in versions 4.7 through 7.1.3. Researcher Umut Bayram of Picus Security discovered that the updater unpacks the new package using the site’s existing, vulnerable code, creating a temporary wp‑content/upgrade folder that remains web‑accessible. An attacker who predicts the folder name can execute code, modify wp‑config.php and overwrite the live site before WordPress deletes the temporary copy. The fix, released in version [...]

ZeroTier announced a partnership with Redapt to embed its ZeroTier Quantum platform into Redapt’s infrastructure, cloud and managed‑service portfolio. The solution uses post‑quantum cryptographic standards ML‑KEM‑1024 (FIPS 203) and ML‑DSA‑87 (FIPS 204) to protect connectivity across data‑center, cloud and edge environments. Redapt will assess, design, deploy and manage the platform for enterprise clients while also providing cybersecurity advisory and virtual CISO services to ZeroTier. Both firms target sectors such as government, defense, finance and regulated gaming, offering a turnkey [...]

Security researchers disclosed a two‑stage attack against Barracuda’s Email Security Gateway appliance that grants unauthenticated attackers root privileges without logging in. The chain combines CVE‑2026‑116606, an authorization bypass that lets a remote user restore a crafted configuration backup and alter global settings, with CVE‑2026‑111560, a command‑injection flaw that normally requires an admin account but is enabled by the bypass. Together they allow arbitrary root‑level commands on the gateway. Barracuda issued firmware patches on Monday, fixing the gateway [...]

Researchers at QiAnXin’s XLab discovered a backdoor embedded in the Control Web Panel (CWP) web server, a Linux hosting management tool. The malicious code, dubbed “cwpdoor,” replaces the legitimate cwpsrv binary for versions 1.15.10 and 1.24.0 and executes attacker‑supplied commands delivered via web requests that bypass access logs. Installation requires root privileges, indicating the attackers have already compromised the host. The backdoor masquerades as an update, restores file timestamps, and evades detection, showing zero hits on VirusTotal’s [...]

QNu Labs has launched a 12‑hour corporate training programme through its QNu Academy, aimed at enterprise cybersecurity leaders. Backed by India’s National Quantum Mission, the three‑day course covers quantum key distribution, quantum random number generators, post‑quantum cryptography, cryptographic inventory management and emerging quantum threats. It targets CISOs, security strategists, risk and compliance officers, IT architects and senior security managers, offering assessment‑based certification and optional on‑site workshops for 15‑50 participants or half‑day executive briefings. Instructors include QNu engineers [...]

Blockchain analytics firm Chainalysis disclosed that a sophisticated agentic cyber‑attack uncovered the names, usernames, password hashes and live tokens of every user on its platform. While the breach prompted the immediate shutdown of its Reactor, Data Solutions, Know‑Your‑Transaction (KYT) UI, API and address‑screening services, the company said no evidence shows that investigation data or attribution records were accessed or altered. Chainalysis has locked affected accounts, halted password resets and is requiring users to reset passwords and rotate [...]

Max Justicz, CTO of Charge Robotics, used OpenAI’s GPT‑5.6 Sol to uncover a severe vulnerability in Ubuntu’s code that could let attackers seize control of servers. The flaw, identified in two hours, was reported to the Debian Project, which promptly issued a patch. OpenAI announced a new cybersecurity‑focused model, ChatGPT Cyber, for its Daybreak Red program, noting it found 948 Linux kernel issues and 81 novel Firefox bugs. The company also pledged a $1 billion initiative to support critical‑infrastructure [...]

Cybersecurity researchers exposed a vulnerability that lets attackers bypass OpenAI’s ChatGPT Lockdown Mode, enabling silent data theft, phishing, and malware delivery. The flaw originates from HTML previews in code blocks that render within nested sandboxed frames. Attackers can escape the sandbox via pop‑ups, gaining unrestricted access to chat content and the user’s session. OpenAI patched the issue on September 18, 2026, but the discovery underscores that AI security hinges on the broader application environment.