Tech Times

npm Supply Chain Attack: North Korea Backdoored 144 AI Packages in 88 Minutes

Mastra AI’s 144 JavaScript packages was executed in just 88 minutes by North Korea’s Sapphire Sleet hacking group, which ...

Copilot searched your mailbox. LiteLLM handed out admin keys. Run this 5-check audit before your stack is next

SearchLeak and a three-CVE LiteLLM chain broke the same AI trust boundary in two weeks. A 5-check audit maps each gap to a ...

The Agentic Reckoning: Enterprise AI organizations have a runtime problem, not a model problem — and most are building the wrong solution

VentureBeat surveyed 132 enterprise AI leaders: the production failure point isn't the model — it's the runtime layer most ...
GitHub

Router-weighted Expert Activation Pruning (REAP)

This repository contains code required to reproduce the expert pruning and merging methods used in the paper: REAP the Experts: Why Pruning Prevails for One-Shot MoE compression Expert pruning and ...
GitHub

Tracing and Attributing Errors in Large Language Model Memory Systems

MemTrace helps developers understand why an LLM memory system gives a wrong answer. A memory system may read many user messages, extract facts, update stored memories, delete outdated memories, ...