Google Project Zero has examined a bypass of Administrator Protection, a Windows 11 25H2 feature intended to replace UAC with more controlled elevation. The finding makes privilege boundaries, workstation hardening, and least-privilege testing immediate concerns for engineering teams.

Google Project Zero says its Pixel 9 exploit chain moved from a zero-click context to Android root using only two exploits. The Dolby vulnerability affected Android broadly until its January 2026 patch, and the researchers are now examining whether a comparable chain can be built for Pixel 10.

Anthropic has opened a research preview of the Model Hardware Standard to an initial group of scientific labs and advanced manufacturers. The proposal aims to give AI agents a shared specification for safely operating physical devices, although its technical design and governance remain to be established publicly.

Identity security is moving beyond periodic access reviews toward continuous discovery, centralized reporting, and behavioral detection. Guidance from AWS, Wiz, and Qualys suggests that access governance, identity-source migration, and attack monitoring should be managed as one lifecycle.

AI agents are being applied to security work that takes action: testing codebases, exploiting vulnerabilities, assessing blast radius, and coordinating response. Cases from AWS, HackerOne, Wiz, and Cisco Talos illustrate the potential for speed—and the need for strict controls over identity, evidence, and execution.

Reports involving malicious Rust crates and exploitable VS Code extension behavior show that supply-chain risk extends beyond production dependencies. Editors, build systems, and plugin ecosystems belong in the same security model as application packages.

Personal repositories can hold work-related code and secrets outside organizational controls. The risk is not solved by assuming every personal repo is harmless or equally dangerous.

Reported malicious arrayref releases executed a backdoor during compilation. For Rust teams, dependency security must cover what build tooling executes, not only production runtime code.

Editor extensions auto-update and run with developer privileges, yet they often fall outside software inventory. The Markdown Preview Enhanced research shows why they belong in supply-chain security governance.

The keyv/cacheable investigation is a reminder that a compromised npm dependency can spread through transitive resolution, not just direct imports. Teams need to scope exposure from build evidence and deployed artifacts.

Wiz argues that AI changes the context around data risk: organizations need to understand what data is connected, exposed, and why. That understanding is a prerequisite for trustworthy agents, response, and remediation.

Wiz says Wiz Workflows is generally available and Remediation and Response is in public preview. The direction moves security beyond isolated alerts toward detection, decision, execution, and verification.