North Korean threat actor UNC1069 deployed seven unique malware families against cryptocurrency targets using AI-generated deepfakes and sophisticated social engineering tactics. The attack involved compromised Telegram …
A deep dive into securing AI agent fleets using just-in-time JWT tokens, role-based access control, and zero-credential architecture. The system includes a meta-workflow that analyzes logs and creates self-improvement …
A comprehensive exploration of how private data can be extracted from AI systems through various attack vectors including fine-tuned models, RAG systems, and vector embeddings. Demonstrates practical attacks against LLMs …
An autonomous offensive security agent discovered critical SQL injection vulnerabilities in McKinsey’s internal AI platform Lilli, gaining full database access to 46.5 million chat messages and compromising the …
An analysis of StrongDM’s fully automated “Software Factory” where AI agents write, test, and deploy production code without human review, exploring the regulatory, accountability, and trust …
This research paper presents six principled design patterns for building AI agents with provable resistance to prompt injection attacks, demonstrating their practical applicability through ten case studies across diverse …
Google Threat Intelligence Group reports on the increasing integration of AI by state-sponsored threat actors for reconnaissance, social engineering, and malware development, including model extraction attacks and …
This research paper presents a domain-specific AI content safety taxonomy for financial services and demonstrates that general-purpose guardrail systems fail to identify most domain-specific risks. The authors evaluate …
AI Police Reports: Year In Review Law enforcement agencies rapidly adopted AI tools for writing police reports in 2024, raising serious concerns about accuracy, transparency, and accountability in criminal justice …
AI-Assisted Static Analysis Uncovers Potential Issues in Curl: Insights from Hacker News A recent Hacker News discussion reveals how AI tools successfully identified legitimate security issues in the curl library, …
