Researcher Details Critical Prompt-Injection Risk in ChatGPT GitHub Workflows
A researcher says hostile instructions hidden in a public repository could redirect an AI-assisted GitHub workflow toward a private repository and expose secrets. The reported 9.3-severity scenario highlights...
Burger King Japan App Breach Exposes 11.18 Million Member Records
Burger King Japan says an intruder accessed the system behind its official app, exposing roughly 11.18 million member records. Payment card data was not held in the affected...
PwnEscape: Linux Kernel Bug Lets Rogue VMs Reach Into the Host
A Linux kernel flaw dubbed PwnEscape (CVE-2026-99999) lets an attacker-controlled virtual machine leak host memory through a checksum-handling bug in the networking stack, with researchers demonstrating a path...
AdFection Campaign Used Online Ads to Deliver Zero-Click Spyware
Citizen Lab researchers have documented AdFection, a zero-click delivery method that used advertising infrastructure to target devices with Candiru-linked spyware. At least 45 Al Jazeera devices were targeted,...
ClickFix Campaign Deploys Stealthy Python Tunnels Ahead of Ransomware Attacks
A investigated Windows intrusion used ClickFix social engineering, DLL sideloading and encrypted Python tunnels to establish durable remote access. Researchers stopped the attack before encryption, but its tooling...
One Click, Full Corruption: Critical Adobe Acrobat Flaw Found by AI-Assisted Fuzzing
Adobe has patched CVE-2026-81983, a critical integer-overflow bug in Acrobat's image-parsing pipeline that lets a malicious PDF corrupt memory the moment it's opened. The flaw was uncovered using...
GitLab Patches Maximum-Severity RCE Flaw as Exploit Code Goes Public
A public proof-of-concept now exists for CVE-2026-97229, a maximum-severity (CVSS 10.0) GitLab vulnerability that chains a CI/CD regex integer overflow into remote command execution. Self-managed GitLab instances on...
Inside Iron Man System: The Phishing-as-a-Service Platform Behind a Billion Fake Emails
Researchers have linked a Chinese-run phishing platform called Iron Man System to roughly one billion phishing emails targeting 75,000 organizations, powered by a library of over 2,300 brand-impersonation...