OpenAI has reportedly received approval from the U.S. Department of Commerce for a broad public launch of its advanced GPT-5.6 model, marking a significant moment in how Washington regulates access to frontier AI systems. A source familiar with the matter…
Tag: Cyber Security News
Accenture Data Breach – Hackers Allegedly Claim to Have Stolen 35 GB of Source Code
A threat actor operating under the alias “888” has posted a listing on a cybercrime forum claiming to sell data allegedly stolen from the IT services and consulting giant Accenture, with the stated haul totaling roughly 35 GB of source…
Anthropic Extends Free Access to Claude Fable 5 on All Paid Plans
Anthropic has extended promotional access to its newest AI model, Claude Fable 5, allowing subscribers on paid plans to use it at no additional cost through July 12, 2026, at 11:59:59 PM PT. The company confirmed the extension in an…
VECT and TeamPCP Reverse Ransomware Kill Chain With Supply Chain Credential Theft
A ransomware strain called VECT has formed an unusual supply chain partnership with a threat group known as TeamPCP, quietly exposing thousands of organizations to compromise before any ransom note appears. VECT’s operators buy their way in using stolen credentials…
The Gentlemen Ransomware With Custom EDR/AV Killers Scaling Faster to Attack Industries Worldwide
The Gentlemen is one of the most rapidly escalating ransomware threats observed in 2026. Emerging in mid-2025 from a payment dispute within the Qilin RaaS program, the group has evolved into a full-spectrum, human-operated Ransomware-as-a-Service (RaaS) operation that Microsoft tracks…
Critical Vulnerability in GCP Dialogflow Allows Attackers to Inject Malicious Code
A critical vulnerability in Google Cloud Platform’s Dialogflow CX that lets attackers inject persistent malicious code into an organization’s AI-powered chatbot pipeline. The flaw, dubbed “Rogue Agent,” disclosed by Varonis Threat Labs, could silently exfiltrate conversations and enable large-scale phishing…
Ubiquiti Disclosed 25 Security Vulnerabilities Across the UniFi Ecosystem
Ubiquiti has disclosed 25 security vulnerabilities affecting its UniFi ecosystem in Security Advisory Bulletin 066, including several critical flaws rated 9.9 and 10.0 on the CVSS v3.1 scale that could allow network-based attackers to fully compromise devices. The advisory spans…
AnyDesk Phishing Attack Uses Scheduled Task Persistence and Artifact Deletion to Evade Detection
A new phishing campaign is turning a trusted remote access tool into a long term backdoor for espionage. Attackers hide behind fake invoices to slip past email filters, and once inside a network they rely on everyday IT software rather…
GitLost Vulnerability Tricks GitHub’s AI Agent into Leaking Private Repos
A newly disclosed vulnerability dubbed “GitLost” shows how attackers can weaponize a single GitHub Issue to trick GitHub’s new AI-powered Agentic Workflows into leaking private repository contents to the public internet, with no credentials, coding skills, or system access required.…
The $50K-to-$5M Gap: Why Your SOC SLA Is Stuck in 2019 — Here’s the 2026 AI SLA Vendor Guide
A technical breakdown of why legacy MDR contract language fails in the age of AI-driven security operations — and the measurable standards that should replace it. The Problem: Contracts Written for Human Queues Pull out your current MDR contract and…
4,982 Security Issues Identified Across 2,259 Affected in Public MCP Servers
A sweeping security crisis across public Model Context Protocol (MCP) servers, cataloging 4,982 security issues across 2,259 affected servers, exposing serious gaps that directly threaten the emerging agentic AI ecosystem. Model Context Protocol has become the dominant standard for connecting…
Critical PHP PDO Driver Bugs Expose Firebird SQL Injection and PostgreSQL DoS Risks
A newly disclosed pair of flaws in PHP’s database driver layer shows that even mature code can hide dangerous surprises. The bugs live inside PHP Data Objects (PDO), the abstraction layer web applications use to talk to databases like Firebird…
U.S. Cyber Defense Agency Reportedly Using Anthropic’s Mythos to Audit Government Code Repositories
The U.S. Cybersecurity and Infrastructure Security Agency (CISA) is reportedly deploying Anthropic’s advanced AI model, Mythos, to audit federal government code repositories, signaling a growing reliance on artificial intelligence for proactive vulnerability discovery. According to Reuters, the initiative, CISA’s Attack…
Windows Adds Microsoft Execution Containers to Secure AI Agent Workflows
Microsoft has introduced Microsoft Execution Containers (MXC), a new security capability designed to protect AI agent workflows on Windows, marking a significant step toward making Windows more trustworthy for autonomous systems. AI agents are rapidly evolving beyond simple assistants into…
Hackers Use Recruiter Phishing Emails and Fake Career Pages to Harvest Gmail Logins
A new phishing campaign is targeting job seekers by posing as recruiters from recognizable brands. The scheme uses fake career pages and worded emails to trick people into handing over Gmail login credentials. What makes this campaign notable is not…
Microsoft Confirms Windows 11, 26H2 Comes With Change in Backup Policy
Microsoft has confirmed a significant change to its Windows settings backup policy with the upcoming release of Windows 11 version 26H2, marking a shift toward improved system resilience and recovery capabilities. According to Microsoft, the Windows settings backup policy will…
OpenAI Codex Desktop App for macOS Vulnerability Allows Attackers to Inject Indirect Prompt
A newly disclosed vulnerability in the OpenAI Codex desktop application for macOS could allow attackers to exploit indirect prompt injection techniques to exfiltrate sensitive data, according to a recent entry in the GitHub Advisory Database. Tracked as CVE-2026-14898, the issue…
Hackers Leverage Microsoft Teams Call to Install RMM Tools and Deploy EtherRAT
Threat actors are now weaponizing something as ordinary as a Microsoft Teams call to slip past corporate defenses and plant a stealthy new remote access trojan called EtherRAT. The campaign blends social engineering with legitimate remote support software, making it…
Cavern Manticore Abuses SysAid RMM and WinDirStat DLL Sideloading to Deploy C2 Framework
A new Iranian-linked hacking group has been caught abusing everyday IT tools to slip malware onto Israeli networks. Researchers have named the group Cavern Manticore, and its latest campaign shows how creative attackers have become at hiding in plain sight.…
16-Year-Old Linux KVM Vulnerability Allows Malicious Guest to Corrupt Host Kernel Memory
A newly disclosed Linux Kernel-based Virtual Machine (KVM) vulnerability, tracked as CVE-2026-53359 and dubbed “Januscape,” exposes a critical flaw that allows a malicious guest to corrupt host kernel memory, breaking the fundamental isolation guarantees of virtualization. The issue, which remained…