AI & LLM
AI and LLM covers generative models, agents, prompt injection, data security and new artificial intelligence tools. The collection connects capabilities, limitations, operational risks and their impact on technical work.
OpenAI Codex: Reported Sandbox Escape Disclosed (ZDI-26-305)
A reported sandbox escape in OpenAI Codex (ZDI-26-305) could potentially allow code execution via specific JavaScript repositories. Th…
Nimbus Manticore: Iranian APT Leverages AI-Assisted Backdoors to Target Aviation and Software Sectors
The Iranian threat group Nimbus Manticore has expanded its operations, targeting aviation and software entities across Saudi Arabia, A…
India’s CERT-In Mandates 12-Hour Patch Window to Counter AI-Driven Exploitation
A new 38-page blueprint from CERT-In slashes the remediation window to just 12 hours for exposed systems, citing the rapid weaponizati…
300 WordPress Zero-Days in 72 Hours for $20: The Falling Economic Threshold of the Bug
TrendAI and CHT Security researchers have uncovered over 300 critical zero-day vulnerabilities in 72 hours using an AI pipeline develo…
NGate Malware Trojanizes HandyPay App to Steal Contactless PINs in Brazil
ESET Research has uncovered a new NGate variant that trojanizes the legitimate HandyPay Android app to relay NFC data and intercept PI…
Fake Data Breach Alerts: When the Warning Becomes the Trap
Cybercriminals are weaponizing 'breach fatigue' and generative AI to craft hyper-realistic phishing alerts that mimic official inciden…
Anthropic’s Project Glasswing Unearths 10,000 Flaws, Triggering 'Patching Paralysis'
Project Glasswing identified over 10,000 critical vulnerabilities in just one month. As Anthropic’s Claude Mythos model accelerates di…
AI Unearths 300 WordPress Zero-Days for $20 Each: The Human Triage Crisis
A high-efficiency AI pipeline has discovered over 300 critical zero-day vulnerabilities in WordPress plugins at an estimated cost of $…
GitLab 19.0 Debuts Native Secrets Management and Air-Gapped AI
GitLab 19.0 integrates native secrets management, agentic merge request workflows, and self-hosted AI models, reinforcing its 'single…
30-Minute Lateral Breakouts: Why the SOC is Losing the Race Against AI-Driven Threats
Average breakout times have accelerated by 29%, with the fastest recorded exfiltration dropping from over four hours to just six minut…
Unit 42: Frontier AI Models Exploiting Open-Source Transparency to Automate Supply Chain Attacks
Frontier AI models are demonstrating the autonomous reasoning required to identify vulnerabilities in open-source code and orchestrate…
Talos Unveils AI Honeypots to Trap Malicious Agents: The Rise of Cognitive Warfare
Cisco Talos demonstrates how generative honeypots can deceive automated AI threats by weaponizing their lack of contextual awareness a…