Petronella Technology Group Blog

Australia To Put Environmental Brakes On AI Data Centers

ISMG Editors: AI Phishing Kits Go Mainstream

New GoSerpent Malware Targets Southeast Asian Governments and Diplomats for Espionage

Zoom Patches Critical Windows Flaw That Could Enable Account Takeover

Nigeria Deepens Cybersecurity Efforts as Cybercriminals See More Profits

Pentagon Suspends CMMC Phase 2 as It Rethinks Contractor Cybersecurity Rules

iCagenda and Balbooa Forms Joomla Flaws Reportedly Exploited as Zero-Days

Hackers Weaponize Balochistan Police Portal in Multi-Group Espionage Campaigns

New Ransomware Exploits Malicious Driver to Remove Cybersecurity Protections

Former ransomware negotiator gets 4 years for BlackCat attacks

Accenture Confirms Data Breach After Hacker Claims Source Code Theft

CISA orders feds to patch max severity ColdFusion flaw by Friday

AI Agent Conducts First Fully Autonomous Ransomware Attack

New Java-Based QuimaRAT MaaS Built to Run on Windows, Linux, and macOS

Potential session/cache leakage between workspace instances or consumer accounts

The Highest-Quality Open AI Model We Have Tested Just Became Fast Enough to Run In-House

Petronella Technology Group, Inc. measured the strongest open-weight AI model in its benchmark program running at production speed on in-house HBM-class GPU hardware. What that means for law firms and regulated organizations that cannot send privileged data to the cloud.

On-Premises AI Coding Assistants Now Rival the Cloud

Petronella Technology Group, Inc. benchmarked open-weight AI coding models on its own hardware against SWE-bench Verified. A self-hosted model resolved 67.3 percent of real issues. What it means for firms handling regulated data.