Cybersecurity News | Daily Recap [12 Sep 2026]

Daily Recap, AI abuse escalated as attackers leveraged Anthropic’s Claude for large-scale distillation and automation, while OpenAI agents were reported to have delivered 2,000+ malicious RubyGem packages to exploit an API key flaw. Meanwhile, high-severity exploitation accelerated with GitLab’s critical file-read issue being probed within a day, and phishing campaigns targeting Microsoft 365 used passkey-themed lures to drive credential theft. #Claude #Anthropic #OpenAI #RubyGem #RubyGems #GitLab #CheckPoint #DutchNCSC #Artifactory #BlueMoon #Chrome #Windows #Microsoft365 #MicrosoftM365 #Houthi #Florida #DMV #BlueMoon

Anthropic says 7 China-based AI labs ran industrial-scale Claude distillation attacks, while attackers also used Claude to automate exploitation, steal data, and even extract secrets from 1.8M Android apps. – AI Distill, Claude Abuse, Android Secrets
Users in Houthi-held Yemen reportedly tried to develop advanced weapons with AI, highlighting the growing misuse of trusted AI platforms for offensive purposes. – AI Weapons
Researchers found OpenAI agents behind a May hacking campaign that uploaded more than 2,000 malicious RubyGem packages and tried to exploit an API key flaw, showing how AI tools can be turned into an attack surface. – RubyGems Hack
Threat actors are increasingly abusing trusted AI platforms for reconnaissance, exploitation, and automation, making them a new attack surface for defenders. – AI Surface

GitLab‘s critical file-read flaw is already being probed across the internet and was exploited within 1 day of disclosure, underscoring rapid weaponization of high-severity bugs. – GitLab Scan, GitLab Exploit, GitLab Probes
Check Point VPN flaws are considered imminently exploitable by the Dutch NCSC, prompting urgent patching guidance for exposed systems. – VPN Flaws
Artifactory flaws were chained in attacks that deployed backdoor malware, showing continued targeting of enterprise software supply-chain components. – Artifactory Chain
BlueMoon exploit kit chained recent Chrome and Windows zero-days, illustrating how multiple fresh vulnerabilities are being combined in active exploitation. – BlueMoon Kit

Passkey-themed phishing campaigns are being used to steal Microsoft 365 data, with attackers abusing trusted authentication themes to bypass user skepticism. – Passkey Phish
New research suggests conventional security awareness testing may miss real-world phishing behavior, reinforcing the need for more realistic training and detection methods. – Phishing Research

Florida confirmed its motor vehicle database was breached via a stolen police account tied to credentials taken from an officer’s personal device, raising concerns about credential hygiene and access control. – Florida Breach, DMV Breach

U.S. airline rules mean passengers affected by a cyberattack-related delay still may not be entitled to a hotel or meal, highlighting gaps in consumer protection during cyber incidents. – Airline Rules

SHARE THIS STORY