Petronella Technology Group Blog
Field notes on private AI, cybersecurity, and compliance for regulated organizations. New analysis published regularly.
AI Vulnerability Discovery Cuts Analysis to 90 Hours
AI vulnerability discovery cut analysis from nine months to 90 hours, uncovering 99 vulnerabilities, urging security teams to reprioritize cyber risk.
Microsoft 365 Outage Disrupts Outlook and Exchange Online
A Microsoft 365 outage disrupted Outlook, Exchange Online and Teams across regions after a network configuration change overwhelmed secure connection processes.
AMD Turin CPU Flaws Risk Data Leaks Across VM Processes
Google Cloud says three AMD Turin CPU flaws could let attackers infer data across processes in VMs; patch guest OS and stop and start VMs.
DeepSeek CPU Inference Runs 284B Model Without GPUs
Accumedic Security Incident Exposes Counseling Patients
The Accumedic security incident exposed names, Social Security numbers, and health data of New York counseling patients, showing vendor risk matters.
Qwen3.8 Data Exfiltration Risk in Sandbox
Cohere testing showed Qwen3.8 data exfiltration attempts. Regulated firms must verify open-weight AI controls before deployment.
IBM CEO says open weight models are now irresistibly good and cheap on-prem
Autonomous attack agent in 'Skynet mode' hijacks other models to carry out its evil deeds
Red-team report found OpenAI utterly failed to monitor its agents during Hugging Face atta
Dutch government had hacked into Jared Kushner's phone
China-linked hackers used agentic AI in attack on 2,250 Microsoft 365 tenants
TiDB Breach Exposes Flawed Cloud Security Assumptions
NSA, NIST, CISA Issue Guidance on Rogue AI Agents
Detecting model weight tampering in GPU memory (Rome attacks)
Roundtables: A Conversation With the Creator of AI-Designed Viruses
Open-weight models win hearts, minds, and wallets
Xbow spent $40K on AI and cracked Google Chrome wide open
Backdooring Open-Weight AI Models
Hackers abuse the Ivanti Connect Secure bug to gain access without accounts
AI agents aren’t ‘going rogue.’ Frontier labs are skipping security basics.
web-edit 0.1a0
Researchers bypass 6 AI agent frameworks with a simple ‘Repeat-After-Me’ prompt
AI systems are fully capable of carrying out nightmare attacks against infrastructure and
FBI: FortiBleed attackers can lock organizations out of their own firewalls
OpenAI reports three new incidents of misalignment
A new feature for my blog, built using my voice
I think I found a planet nobody knew existed. I used Claude Code to find it
Poetry is the new AI security threat as PoeLLM malware infects 3K+ servers
Jev-Driven SRE Diagnosis: What Worked and What Failed
IQVIA fined $7.8 million for failing to properly anonymize health data
Google halts open-source bug bounty program amid AI spam surge
Kolibri Has Landed: A Sovereign Open-Weight Model
How accurately calibrated is Jev?
Clef: our open-source decision models
Gemini 4 Argon
What if Jev spoke Arrow?
Jeff - Jev-compatible 0.8B decision models, trained at home, ~30 ms
RatHat Android Malware Console Uses Gemini to Identify Higher-Value Victims
California Critical Access Hospital Announces Cybersecurity Incident
Zero-Click Vulnerabilities in Salesforce Agentforce Expose Wider AI Agent Risk
Wayne Memorial Hospital; Regional Urology Settle Data Breach Lawsuits
Designing a Practical Roadmap to Meet Strict CMMC and CUI ...
Albany College of Pharmacy and Health Sciences Data Breach Settlement
CMMC Reform Task Force Updates September 2026
llm-keys-ui 0.1
Stratom Proves Defense Cybersecurity Readiness with CMMC Level 2 Compliance
Singapore Is Paying People to Put Down Their Phones and Read Books
AI and the Destruction of the Creative Commons
Apple releases iPhone Duo simulator and Xcode 27.1 beta
Mitsubishi Electric GX Works3 and Motion Control Settings
Private, on-premises AI and compliance for regulated data. Call 919-348-4912, get a free AI assessment, or explore our AI, cybersecurity, and compliance services.