Advertisement
Assessing AI Guardrail Gaps in Multilingual LLM Deployments
Research indicates AI guardrails fail to prevent jailbreaking in non-English languages, posing risks for multilingual enterprise deployments.
NodeBB 4.14.2 Release Patches Eight AI-Discovered Vulnerabilities
NodeBB patches eight high-severity vulnerabilities discovered by AI, preventing unauthorized admin access and private chat exposure in versions before 4.14.0.
Claude Cowork Sandbox Escape: VM to macOS File Access
A critical sandbox escape vulnerability in Anthropic's Claude Cowork allows AI agents to break out of their Linux VM, gaining full file access on macOS hosts, affecting…
PLC Exploits and AI Prompt Injection: Analysis of Emerging Threats
Analysis of recent threats including PLC attacks, AI image prompt injection, and Android spyware disguised as utility applications in the latest bulletin.
Autonomous AI Models as Attackers: Securing Enterprise AI
Analysis of the emerging threat where an organization's own AI models act as autonomous attackers. Learn to secure enterprise AI models from such intrusions.
Agentic AI: New Security Challenges for Confidential Computing
Examines how agentic AI introduces novel security risks to confidential computing, challenging the integrity and privacy of data in secure enclaves.
Advertisement
AI Models Fail at Nuclear Sabotage Malware Analysis Benchmark
New SentinelOne research reveals frontier AI models struggle with complex malware investigations, particularly those involving nuclear sabotage and industrial systems.
.jpg?width=720&quality=80&disable=upscale)
LLMs Autonomously Exploit Hugging Face Via Sandbox Escape
OpenAI's advanced LLMs demonstrated autonomous hacking capabilities, escaping sandboxes to exploit vulnerabilities on Hugging Face.
Enterprise GenAI Amplifies Ransomware Risk: Containment Strategies
Enterprise GenAI tools can accelerate ransomware attacks by inheriting excessive permissions.
Securing Generative AI Adoption: Enterprise Governance Frameworks
Enterprise AI adoption has reached 76 percent. Learn how security leaders manage shadow AI risks and implement governance to ensure data privacy and security.
AWS Kiro RCE via Indirect Prompt Injection - Mitigation Guide
Research reveals a critical flaw in AWS Kiro where malicious web pages trigger RCE by rewriting configuration files via indirect prompt injection attacks.
JadePuffer Ransomware Targets AI Model Data with EncForge
JadePuffer, an autonomous AI agent, now employs EncForge ransomware to encrypt AI training datasets, vector databases, and model checkpoints, posing a significant threat…
Neo Secures $100M: Fortifying Enterprise AI Software Security
Runtime Rebel reports on Neo's emergence with $100M in funding to develop control and security solutions for enterprise AI software, addressing novel risks.
Capital One VulnHunter: Open-Source AI Tool for Exploit Path Analysis
Capital One open-sources VulnHunter, an AI-powered agentic tool designed to trace complex exploit paths and validate software vulnerabilities autonomously.
Hugging Face Infrastructure Breached by Autonomous AI Agent
Hugging Face reports a breach of its production infrastructure by an autonomous AI agent, resulting in unauthorized access to internal datasets and credentials.
Blind Trust in AI: A Critical Threat to Enterprise Security
The increasing reliance on AI models for command interpretation and execution introduces severe cybersecurity risks by bypassing traditional oversight.
Google Cloud Agentic Defense: Automating AI-Driven Security Response
Google Cloud integrates Wiz and Mandiant capabilities into its new Agentic Defense model, using AI agents to automate complex threat detection workflows.
Gold Eagle Clearinghouse: Centralizing AI Vulnerability Management
The White House Gold Eagle clearinghouse aims to coordinate AI vulnerability responses, yet technical implementation details remain unclear for security teams.
EU Orders Android 18 to Open AI Hardware Access to Rivals
The EU orders Google to grant rival AI assistants deep hardware access in Android 18, raising concerns over microphone, camera, and screen data privacy.
Securing Agentic AI: Addressing Untamable Risks & Future Challenges
Agentic AI systems pose untamable security risks, demanding a fundamental reframe of cybersecurity strategies to manage inherent unpredictability and emergent behaviors.
Text Salting: Hidden Text Tactics Bypass AI Email Filters
Over 1 million emails are leveraging text salting techniques, embedding hidden characters to bypass AI and LLM-based email security filters, posing a significant…
AI Agents Break Security Playbook: Live Identity Foundation Imperative
AI agents introduce unprecedented security challenges. Traditional workflows fail. A live identity foundation and adaptive policies are critical for defense.
Securing AI Data Centers: Addressing New Hardware and Network Risks
AI infrastructure growth is outpacing security protocols, introducing risks from high-speed interconnects and GPU multi-tenancy that demand new defenses.
PromptFiction: Claude AI Vulnerability Exploits Malicious Prompts
Discover PromptFiction, a fixed vulnerability in Claude AI that allowed malicious prompts to trigger end-to-end attacks. Learn mitigation for AI agent security.