Nvidia has introduced an AI Agent Safety Platform designed to prevent autonomous AI agents from exceeding their designated operational boundaries.
Tag: CyberMaterial
Drunk AI models leak secrets, easier to jailbreak
Researchers at the University of New South Wales (UNSW) Sydney have found that artificial intelligence models trained to write like intoxicated people…
New Mexico Jury Finds Facebook Liable for Privacy Deception
A jury in New Mexico has delivered a verdict against Facebook, finding the social media giant liable for deceiving users about the privacy protections…
Psychedelic Stealer Campaign Targets Ukrainian Businesses
Arctic Wolf researchers have uncovered a malware campaign targeting legitimate Ukrainian businesses with an information stealer called Psychedelic Stealer.
Storm-3168/JADEPUFFER Azure Credential Abuse
Microsoft has released its first comprehensive report on Storm-3168, a threat actor that Sysdig previously identified as JADEPUFFER and linked to what…
Cyber Briefing: 2026.09.25
Ransomware exploitation, phishing, exposed customer data, and failed detection rules are creating opportunities for unauthorized access, credential theft,…
Ransomware gangs exploiting critical TeamCity flaw
The U.S.
Party Invite Phishing Scams Target Users
A wave of phishing attacks is targeting users through fake party invitations that mimic popular digital invitation services such as Evite and Paperless…
CenterPoint Energy breach: 7.49M records claimed stolen
CenterPoint Energy disclosed a data breach on September 14, 2025, confirming that an unauthorized third party obtained customer information through one of…
Rydox marketplace admin pleads guilty
A citizen of Kosovo has pleaded guilty to federal charges for operating Rydox, a significant illegal online marketplace that trafficked in stolen personal…
Detection dashboards mask security coverage gaps
Nearly half of all security detection rules deployed across enterprise environments are failing to function properly, according to new research from…
Microsoft Unified Copilot App Launches with Code, Autopilot
Microsoft released a unified Copilot application today that consolidates its previously scattered consumer and enterprise AI tools into a single client.
Cyber Briefing: 2026.09.24
Critical SolarWinds flaws and compromised MemTensor packages create opportunities for remote code execution and credential theft, while an AI-agent breach…
Data Quality Overtakes Skills as Top Threat Hunting Challenge
Data quality has emerged as the primary obstacle facing threat hunting teams, overtaking skills shortages for the first time in five years of tracking by…
NIST Updates OT Security Guide; CISA/FBI ICS Advice
The National Institute of Standards and Technology has opened Revision 4 of its operational technology security guide for public comment, with the…
SolarWinds Patches Critical RCE Flaws
SolarWinds has issued security patches addressing two critical remote code execution vulnerabilities in its Observability Self-Hosted platform.
OpenAI Agent Hacked Australia Health Service
A security breach involving an OpenAI agent has compromised an Australian health service, triggering a formal government investigation into whether the…
Revolut launches facial recognition checkout system
Financial technology company Revolut has launched a pilot program for facial recognition payments at retail locations, starting with a three-day trial at…
MemTensor npm/PyPI packages compromised
The popular MemOS framework for large language models suffered a supply chain attack early Wednesday when attackers published malicious versions…
Cyber Briefing: 2026.09.23
AI-enabled phishing, browser vulnerabilities, compromised third-party credentials, and exposed customer data remain key areas of concern, with EvilTokens…