Mozilla researchers revealed a new attack that tricks Claude Code into running hidden commands from seemingly harmless GitHub ...

Attackers can inject indirect prompts in normal-looking repositories to trick Claude Code into spawning a reverse shell.

Sysdig says JADEPUFFER used CVE-2025-3248 in Langflow to automate intrusion, credential theft, encryption, and data wipe.

NVIDIA diffusion language model Nemotron TwoTower achieves 2.42x LLM inference throughput without a full retraining run, ...

Princeton’s CEO-Bench gave 14 AI models $1 million to run a simulated SaaS startup for 500 days. Most went bankrupt or lost ...

New benchmarks show semantic code graphs helping coding agents find change locations faster and complete updates more ...

A new framework called SkillWeaver tackles AI agent tool routing by skipping full-library loading, cutting token use 99% on ...

One condition did what my nagging couldn't ...

Everything you need to know about how we analyzed the 13,000+ comments submitted in the federal government’s request for ...

Meta's new Pocket app lets users create and share interactive mini-games using plain text prompts, making vibe coding ...

Qualcomm, which just a few years ago had no data-center revenue, intends to generate $15 billion in data-center sales by 2029 ...

AI agents are now taking over repetitive work, identifying issues humans may miss, and helping teams maintain testing speed ...