The recent developments surrounding vulnerabilities in major AI repositories like Hugging Face serve as a critical wake-up call for the cybersecurity community. As we accelerate toward an agentic future, the platforms we rely on for innovation are increasingly becoming the primary vectors for systemic risk.
Security operations teams face a constant balancing act: stopping sophisticated email threats, maintaining visibility across their attack surface and keeping administrative workflows running smoothly. When security tools operate in silos or rely on rigid, manual processes, friction builds up quickly. This friction consumes valuable time that analysts could spend on higher-priority initiatives.
Researchers at ReliaQuest are tracking two new phishing toolkits that are designed to bypass multifactor authentication (MFA). The first tool, called βJalisco,β is a device code phishing platform that pairs with AI-powered phishing-as-a-service platforms like EvilTokens to provide fresh OAuth codes in real time.
A new Windows-focused infostealer and remote access trojan (RAT) dubbed Dolphin X is being advertised on cybercrime forums with a clear pitch: automate the theft and triage of high-value corporate targets. Unlike commodity stealers that focus mainly on browser passwords, Dolphin X is positioned as an enterprise-adjacent data vacuum with a built-in AI-powered victim scoring [β¦]
The post New Windows Stealer Uses AI Profiling to Identify High-Value Corporate Victims appeared first on GBHackers Security | #1 Globally Trusted Cyber Security News Platform.
Google has unveiled CodeMender, a managed AI security agent designed to identify, validate, and remediate software vulnerabilities at machine speed. Announced in preview on July 22, 2023, the tool is available through the Gemini Enterprise Agent Platform and can also function as a core component of Googleβs AI Threat Defense offering. This launch comes as [β¦]
The post Google Unveils CodeMender AI Agent for Automated Vulnerability Detection and Remediation appeared first on GBHackers Security | #1 Globally Trusted Cyber Security News Platform.
Anthropic has launched the Claude Security plugin for Claude Code in beta, enhancing its AI-assisted development platform with security scanning capabilities designed to identify vulnerabilities earlier in the software development lifecycle. The company stated that developers can scan code changes before committing them or initiate comprehensive security reviews across an entire codebase directly from the [β¦]
The post Anthropic Launches Claude Security Plugin to Scan Codebases for Vulnerabilities Before Commit appeared first on GBHackers Security | #1 Globally Trusted Cyber Security News Platform.
Threat actors are using a new technique called βphantom squattingβ to trick AI tools into directing users to phishing sites, according to researchers at Palo Alto Networksβ Unit 42.
Since AI models frequently hallucinate phony information, they sometimes point users to websites that donβt exist. Threat actors are now registering these AI-hallucinated domains and using them to host phishing sites.
Cybercrime used to have a β"tell."Β It was the digital equivalent of a villain stroking their cat - clunky grammar, misspelt links and suspicious attachments that screamed βphishingβ.
Yubico has released the YubiKey firmware version 5.8, expanding its hardware security key platform beyond phishing-resistant authentication. This update introduces verifiable, hardware-backed authorization for digital signatures, identity wallets, payment confirmations, and AI agent approval workflows. Announced on July 21, 2026, this firmware update aims to help enterprises verify not only who accesses an application but [β¦]
The post Yubico Launches YubiKey 5.8 With Hardware-Backed Authorization for AI Agent Workflows appeared first on GBHackers Security | #1 Globally Trusted Cyber Security News Platform.
OpenAI has revealed that during an internal evaluation of advanced cyber capabilities, AI agents exploited a zero-day vulnerability, escaped a constrained research environment, and compromised parts of Hugging Faceβs production infrastructure. While Hugging Face detected and contained the activity, OpenAIβs internal security team also identified unusual behavior during the assessment. OpenAI Compromise Hugging Face Servers [β¦]
The post OpenAI Exploits Zero-Day to Gain Internet Access and Compromise Hugging Face Servers appeared first on GBHackers Security | #1 Globally Trusted Cyber Security News Platform.
So, you have some AI tools or are thinking about deploying them and want to know a bit about securing them.
You are not alone, but there are significant challenges due to the rapidly growing capabilities of AI, and the issues around new types of vulnerabilities we may not be used to thinking of. This is a very challenging area to attempt to secure, but I hope to point you in the right direction and set you up with some resources.
JADEPUFFER has escalated from automated database extortion to purpose-built AI model destruction, deploying a custom Go ransomware dubbed ENCFORGE to encrypt and effectively wipe highβvalue AI and ML artifacts across an entire stack. A missingβauthentication bug in the /api/v1/validate/code endpoint that enables unauthenticated arbitrary Python execution on the host. That initial operation chained reconnaissance, credential [β¦]
The post JADEPUFFER Deploys ENCFORGE Ransomware Built to Destroy AI Models and Training Data appeared first on GBHackers Security | #1 Globally Trusted Cyber Security News Platform.
Iran-linked APT42 is escalating its espionage operations with AI-assisted phishing and an expanded TAMECAT backdoor, enabling long-lived access to defense and government identities rather than just endpoints. Recent activity shows tightly integrated social engineering, cloud abuse, and fileless PowerShell tradecraft that significantly complicate detection and response. APT42, also tracked as TA453 in some reporting, is [β¦]
The post Iran-Linked APT42 Uses AI-Assisted Phishing and TAMECAT Backdoor to Target Defense Officials appeared first on GBHackers Security | #1 Globally Trusted Cyber Security News Platform.
AgentBaiting is the clearest sign yet that AI agents and their capability ecosystems have become a firstβclass malware delivery surface, with FakeGitβs 7,600βrepo operation pushing SmartLoader and StealC directly into AI Skills and MCP workflows. By turning agentβreadable READMEs, public AI registries, and GitHub trust signals into a weaponized βAI capability supply chain,β attackers now [β¦]
The post AgentBaiting Uses Fake AI Skills and MCP Servers to Deliver SmartLoader and StealC Malware appeared first on GBHackers Security | #1 Globally Trusted Cyber Security News Platform.
Bit2Watt is a newly disclosed cyberβphysical attack class that weaponizes AI and GPU workloads in modern data centers to destabilize nearby power grids, turning compute infrastructure itself into a gridβscale threat surface. Measurements on NVIDIA accelerators show subβmillisecond power ramps where a single Volta V100 or RTXβseries GPU swings from low-load phases to nearβTDP draw, [β¦]
The post Bit2Watt Attack Turns AI Data Centers Into Cyber-Physical Threats to Local Power Grids appeared first on GBHackers Security | #1 Globally Trusted Cyber Security News Platform.
Regulated enterprises evaluating AI pentesting platforms should assess eight capabilities: FedRAMP Moderate authorization or higher, multi-framework compliance support, human-in-the-loop with agentic AI, verified zero-false-positive findings, bidirectional workflow integrations, self-service launch, auditable coverage visibility, and full offensive security platform capabilities. Vendors who can't clearly distinguish their AI from an automated scanner are likely selling exactly that.
The post The AI Pentesting Platform Checklist for Regulated Enterprises appeared first on Synack.
Login