Follow on Telegram

tech/security

The NSA and the Emergence of AI Cyber Doctrine

The NSA's Quiet Move Into AI Cyber Doctrine Recent reports suggest the NSA (National Security Agency) is developing what some are calling a new "Mythos"—a framework for integrating artificial intelligence into cybersecurity doctrine. This represents a significant evolution in how the agency approaches national cyber

AI-Enabled Sanction Evasion Emerges as New IT Governance Challenge

The Growing Problem of AI-Assisted Sanction Evasion Organizations managing compliance and IT governance are facing a new challenge: artificial intelligence being leveraged to circumvent international sanctions. Security researchers and regulatory bodies have begun documenting cases where bad actors use AI tools to analyze sanctions lists, identify gaps, or craft communications

Illinois Legislature Advances Landmark AI Frontier Model Safety Bill to Governor

Illinois has taken a notable step in AI governance with state lawmakers advancing legislation focused on frontier model safety. The bill, described as landmark by transparency advocates, has cleared the legislative process and now awaits action from Governor Pritzker. The legislation would establish requirements for safety evaluations of advanced AI

House NDAA Proposes Protected Disclosure Program for AI Incidents

The House of Representatives has included a provision in its version of the National Defense Authorization Act (NDAA) that would establish a protected disclosure program for artificial intelligence incidents. The proposed program would create a formal mechanism for reporting AI-related safety concerns and incidents, mirroring similar whistleblower-style protections found in

Research Shows AI Models More Vulnerable to Iterative Attacks Than Previously Understood

New research suggests that AI models can be more vulnerable to iterative attacks than the systems' developers have claimed. The findings challenge assumptions about AI robustness, particularly regarding how models respond to repeated or evolving adversarial attempts. The research highlights that while AI companies often market their models as

Technology News

Huawei Adapts as Moore's Law Era Winds Down

The semiconductor industry is entering a new phase as the long-standing principle known as Moore's Law—that the number of transistors on a chip doubles roughly every two years—faces physical and economic constraints. Huawei, led by figures within the company reportedly dubbed the "Chip Queen,"

FastAPI Authentication Flaw Puts AI Tools at Risk

A security flaw in the Starlette framework has been identified that allows authentication bypass in applications built with FastAPI. This vulnerability affects AI tools and applications relying on FastAPI's authentication mechanisms. The flaw potentially exposes sensitive AI tools to unauthorized access, as the authentication layer can be circumvented.