Cyber

Up to 70% of websites were said to be affected by the PHP2Shell flaw in October 2026, capping coverage increasingly focused on AI-enabled cyber defense and vulnerabilities.

Who they are

Cyber is a product-label cluster used in coverage of cybersecurity: attacks, software vulnerabilities, defensive tools, and the operational failures they can cause. Its stories also include older Cyber Monday retail roundups and Tesla’s Cybercab, but the recent coverage is chiefly about the security consequences of widely deployed technology and the tools built to manage them.

The recent arc

Recent coverage has shifted toward cyber defense in the AI era. The largest recent concentration came in 2026 Q3, led by Anthropic’s reports on disrupting attempted misuse of Claude for cyberattacks, influence operations, and surveillance, alongside OpenAI’s Daybreak tiers and a proposed public listing for cyber-defense company RedLattice. The October report of the PHP2Shell memory-management vulnerability, said to affect up to 70% of websites, extended that focus from AI tooling to software exposure at internet scale.

The prior phase was driven more visibly by disruption and criminal incidents: CrowdStrike’s defective update caused Windows BSODs that took banks, airlines, and other businesses offline in July 2024, while UK police arrests tied to the M&S and Co-op hacks featured in 2025. Coverage also tracked security concerns in public-facing systems, including experts’ criticism of the EU age-verification app’s privacy and security posture.

The tension

The coverage circles a dual-use tension: the same frontier AI systems that Anthropic and OpenAI position for defensive cybersecurity can be misused for attacks, surveillance, and influence operations. It also underscores the fragility of concentrated technology infrastructure, where a CrowdStrike update affecting Microsoft Windows systems or a critical PHP flaw can turn a security or software failure into a broad operational event.

Why it matters

If this trajectory continues, cyber coverage will increasingly be about whether AI providers, software maintainers, and specialist defenders can improve detection and response faster than new attack methods and deployment risks spread. The RedLattice listing plan suggests investor interest in that defensive layer, but the CrowdStrike outage and EU-app concerns show that security products and policies can themselves introduce material reliability and privacy trade-offs.

Related: Crowdstrike · Microsoft · BSODs hit thousands of Windows PCs due to “a defect” in an update from · Anthropic publishes a threat intelligence report on how it disrupted e · U.S.