Claude AI Guided Hackers Toward OT Assets During Water Utility Intrusion
Dragos reported that attackers used Claude and GPT models during an intrusion into a municipal water and drainage utility in Monterrey, Mexico. The AI tools helped the actor plan activity, build tooling, process victim data, and identify OT assets, including a SCADA and IIoT management interface. This matters because it shows AI being used in a practical intrusion workflow against critical infrastructure, not just for phishing or malware generation.
vm2 Node.js Library Vulnerabilities Enable Sandbox Escape and Arbitrary Code Execution
Researchers disclosed a dozen critical vulnerabilities in the vm2 Node.js library that could let attackers escape the sandbox and execute arbitrary code on the host system. vm2 is used to run untrusted JavaScript in restricted environments, so failures in that boundary can become serious application and supply chain risk. Teams using vm2 should confirm affected versions, update quickly, and review where untrusted code execution is allowed in their applications.
CallPhantom Android scam reached 7.3 million downloads on Google Play
ESET researchers found 28 fraudulent Android apps on Google Play that claimed to provide call histories, SMS records, and WhatsApp call logs tied to phone numbers. The apps reportedly reached more than 7.3 million downloads before removal, but instead of retrieving real records, they generated fake data and pushed users toward payments or subscriptions. This is useful for security teams because it shows how convincing mobile scams can scale through official app stores, even without sophisticated malware.
Iranian cyber espionage disguised as a Chaos Ransomware attack
Rapid7 assessed that an intrusion posing as a Chaos ransomware incident was likely an Iranian state-sponsored espionage operation linked to MuddyWater. The attackers used Microsoft Teams social engineering, screen sharing, remote access tools, credential theft, data exfiltration, and extortion-style messaging to make the activity look financially motivated. The practical takeaway is that responders shouldn’t stop at the ransomware narrative, especially when remote access tooling and persistence suggest a longer-term intelligence objective.
A DOD contractor’s API flaw exposed military course data and service member records
Researchers found that an API flaw in Schemata’s AI-powered training platform exposed user records, military course data, organization information, and links to documents stored in AWS. The issue reportedly allowed a low-privilege account to access data across tenants, including service member names, emails, base assignments, and training enrollments. This matters because tenant isolation failures in SaaS and AI platforms can expose sensitive operational context even when the data isn’t classified.
The post InfoSec News Nuggets 05/07/2026 appeared first on AboutDFIR – The Definitive Compendium Project.
