Grivyonx Intelligence Hub
Older cybersecurity, AI and cloud security analysis from the Grivyonx research team.
Fortinet’s FortiGuard Labs uncovered a prolonged supply‑chain breach of QuickFox VPN, where a trojanized Windows installer delivered the FDMTP backdoor. The attack, ongoing since August 2025, highlights the risk of compromised software pipelines.
By Gourav Rajput
In the fast‑moving world of artificial intelligence, trust is built on performance benchmarks but also on rigorous security scrutiny. Independent cybersecurity teams regularly put AI models through their paces to uncover hidden weaknesses before they reach end users. OpenAI recently shared details about a few third‑party evaluation runs that exposed unexpected behaviors, and it outlined a set of new safeguards designed to make future testing more transparent and resilient.
By Gourav Rajput
The commercial phishing‑as‑a‑service toolkit known as Greatness has recently incorporated device code phishing, a tactic that exploits the OAuth 2.0 Device Authorization Grant to bypass multi‑factor authentication. This development marks a significant shift in how threat actors can harvest credentials and tokens without directly confronting traditional security controls.
By Gourav Rajput
A malicious npm worm called Keyv-Linked compromised hundreds of packages, including Claude Code and VS Code. This article explores its spread, impact, and how to protect your projects.
By Gourav Rajput
Cybercriminals are masquerading as legitimate Adobe and Zoom software updates to trick users into installing the ScreenConnect remote management tool. This campaign, dubbed SMOKE#SCREEN, leverages convincing lures to gain persistent remote access. Below we break down the attack chain, technical details, and actionable mitigation steps.
By Gourav Rajput
A recent security glitch in the AI‑powered meeting assistant tl;dv has opened the door for malicious actors to snoop on confidential video calls. The flaw lies in a Firebase misconfiguration that grants unauthorized users access to other participants’ meeting metadata. This breach could potentially expose sensitive discussions involving government agencies and Fortune 500 companies picture a scenario where an attacker can simply query a public endpoint and retrieve details of any ongoing or past meeting.
By Gourav Rajput
The cybersecurity industry has long relied on a hierarchy of attacker sophistication to model risk. That model is crumbling. With the rise of 'vibe hacking,' large language models are acting as force multipliers, granting novice actors the operational capacity of seasoned professionals. This shift demands a fundamental rethink of how we detect, prioritize, and respond to threats.
By Gourav Rajput
Google has removed three AI agent workflows from its Agent Development Kit Python repository after researchers from Pillar Security uncovered a serious security flaw. The vulnerability involved a public GitHub issue that could manipulate a triage agent into activating a privileged code-fixing agent through prompt injection techniques.
By Gourav Rajput
The U.S. Cybersecurity and Infrastructure Security Agency has placed a high-severity vulnerability in N-able N-central on its Known Exploited Vulnerabilities list after confirmed attacks in the wild. The flaw, CVE-2026-18577, exposes managed service providers and their clients to serious security risks due to incomplete remediation of a prior critical issue.
By Gourav Rajput
A Cambodia-based criminal network exploited ChatGPT to fuel investment, romance, and gambling scams. OpenAI intervened, showcasing AI's dual role in enabling and preventing fraud. This case highlights the evolving battle between cybercriminals and tech giants.
By Gourav Rajput
OpenAI has publicly addressed Apple's recent legal action, providing clarification on disputed claims about employee communications and sharing documentation that sheds light on the situation. The tech giant's response comes amid ongoing tensions between the two major players in the artificial intelligence space.
By Gourav Rajput
A team of researchers has developed a groundbreaking tool capable of tracing AI-generated videos back to the specific model or system that created them. The innovation aims to foster collaboration across the technology industry to build stronger defenses against misinformation and digital manipulation.
By Gourav Rajput
A recent incident involving Anthropic's Claude AI highlighted a critical security flaw: over-permissioning, especially unrestricted internet access, not model imperfections. This article dissects the root causes and offers insights into securing AI deployments.
By Gourav Rajput
<p>Recent cybersecurity research has uncovered a dangerous trend: 18 malicious npm packages are being distributed to users of Alibaba developer tools. These packages secretly embed a cross-platform remote access trojan (RAT), enabling attackers to infiltrate systems undetected. The attack highlights the growing sophistication of supply chain threats targeting Chinese-speaking environments.</p><p>One of the most concerning packages is 'lib-mtop,' which mimics a legitimate Alibaba private package but operates as a Trojan horse. Researchers warn that this attack vector could compromise sensitive data and system control across multiple platforms.</p>
By Gourav Rajput
New research from Unit 42 reveals three distinct attack paths targeting Chrome's Google Password Manager that could allow malware to silently hijack passkey-protected accounts. The most severe chain, called Golden Pass-ta-key, compromises the master encryption key itself, bypassing all user-facing authentication prompts.
By Gourav Rajput
A new wave of ransomware attacks has been traced back to the INC Ransomware group, which is aggressively exploiting known vulnerabilities in SonicWall Secure Mobile Access 1000 series appliances. Cybersecurity firm Resecurity has identified a sharp increase in activity since August 2026, with the group listing numerous victims on its public data leak platform. Organizations relying on SonicWall SMA 1000 VPN solutions are urged to patch immediately and review their security posture.
By Gourav Rajput
Researchers intercepted a Chinese‑backed threat actor’s use of the DeepSeek AI model to compromise more than 1,200 hosts. The attackers employed proxyjacking to stealthily move laterally and launch attacks against a leading security firm. The incident highlights the growing danger of AI‑empowered adversaries and underscores the need for advanced detection, rapid response, and robust governance around emerging AI technologies.
By Gourav Rajput
This week’s security landscape highlighted how permissive configurations and overlooked weaknesses can lead to major incidents. From autonomous AI overreach to massive crypto heists, critical utilities, and dangling DNS records, attackers found plenty of open doors.
By Gourav Rajput
