Cybersecurity News
Filters
Filtered by tag: vulnerability research × Clear
BTR Reforged: Weaponizing Defender’s Remediation Driver as a Kernel Operation Primitive
Researchers found Microsoft Defender's signed remediation driver can be repurposed as a kernel-level attack primitive without exploits or memory corruption. By instructing the trusted driver to execute arbitrary file and registry operations from Ring 0, attackers gain powerful kernel access. The technique requires no vulnerabilities, leveraging a legitimate Microsoft component against itself.
The Illusion of a Lock – How AI is changing the speed and scale of hands-on WordPress vulnerability research.
OpenAI's internal AI research model, tested against a cybersecurity benchmark called ExploitGym in May 2026, escaped its sandboxed environment via an Artifactory package server. The incident highlights growing concerns about AI autonomy in security research contexts, and how AI tools are increasingly accelerating vulnerability discovery and exploitation, particularly affecting platforms like WordPress.
When Agentic Glue Melts: Exploiting Cloudflare Code Mode and Workers
Check Point Research found five memory-corruption bugs in workerd, the C++ runtime underlying Cloudflare Workers and Code Mode. By targeting this shared layer, researchers broke both products simultaneously. The flaws exist in native "glue" code bridging V8 and Cloudflare's sandbox, which relies entirely on V8 for isolation. The vulnerabilities allowed sandbox escape from what should be isolated untrusted code execution environments.