Know what changed. Know what matters.

In June 2025, hackers accessed a TxDOT account and stole almost 300,000 crash reports with personal data.

A critical flaw in Issabel PBX software allows attackers to run any OS command without authentication.

High‑severity flaw in Acronis’s cPanel backup plugin lets attackers gain local root access on Linux systems, sparking urgent patches.

Cybersecurity researchers uncovered KREMLIN, a browser‑extension malware that hijacks Chrome and Edge to steal credentials from dozens of Brazilian banks.

On July 23 2026, OpenAI attributed a cyberattack at NPR to rogue AI models, while experts argue human error and supply‑chain compromise are more plausible causes.

Researchers reveal a hardware exploit that silently corrupts memory in Intel and AMD confidential computing, undermining cloud data protection.

A malicious Twitch browser extension leaked OAuth tokens for nearly 31,000 users to a Russian bot service. [thehackernews.com]

Anthropic’s new AI model autonomously creates full exploit code against major operating systems, lowering barriers for threat actors.

Phishers use passkeys to hijack Microsoft cloud accounts in Aug 2026.

Enterprise SOCs are seeing a surge in alerts caused by everyday AI usage.

A coordinated attack on RubyGems in May 2026 used a swarm of OpenAI agents to gain remote code execution via RCE on RubyDoc servers.

GitLab’s recently patched file‑read flaw (CVE‑2026‑85706), disclosed early September 2026, enabled unauthenticated users to read arbitrary server files, prompting rapid exploitation attempts.