CERT-EU warns of rapid rise in attacks driven by agentic AI frameworks such as Artex
The EU institutions' CERT says criminals are abusing autonomous penetration-testing agents to run multistage intrusions faster, at scale and with less expertise.
Insights · Topic
Every briefing we have filed under AI Security — newest first.
91 results in AI Security Clear filters
The EU institutions' CERT says criminals are abusing autonomous penetration-testing agents to run multistage intrusions faster, at scale and with less expertise.
Mittwald says an attacker exploited its Model Context Protocol interface for mStudio on 27 October, forcing a temporary lockout of admin and SSH access for all customers.
Tampered builds 2.1.312 and 2.1.313 of the Claude Code extension were planted in Cursor's marketplace cache for about 11 hours; Cursor later traced the breach to a key exposed by Fireworks AI.
The Spanish data protection authority says an organisation reported a breach in which an autonomous AI agent logged in, hunted for application flaws, altered personal data and reached invoices.
China's national virus response centre found counterfeit OpenClaw skill packages carrying malicious code that downloads trojans and steals user data when an AI agent calls the skill.
Cyera researchers disclosed a CVSS 10.0 content-type confusion bug in the n8n workflow automation platform that can lead to full compromise. About 100,000 servers are estimated to be at risk.
The Berlin Commissioner for Data Protection has officially requested the removal of the DeepSeek AI application from Google Play and Apple App Store, citing violations of the General Data Protection Regulation (GDPR).
Google has announced the rollout of Veo 3, an advanced video generation tool now available to all users of Vertex AI, Google Cloud's machine learning testing platform.
Cybercriminals are increasingly leveraging sophisticated artificial intelligence (AI) models to develop malware and craft phishing scams, as highlighted by recent findings from Cisco Talos.
Facebook, a division of Meta Platforms, is seeking user participation by prompting them to upload images from their mobile devices.
Security operations centers (SOCs) face increasing pressures as threats become more complex and frequent while security budgets remain stagnant.
Cybercriminals are increasingly circumventing the protective measures designed to prevent AI models from being exploited for illicit activities, according to recent research.
Start a conversation
Our team can check your exposure to a specific vulnerability or campaign within hours.