Ai Security

Tailscale found no vulnerability in the Hugging Face intrusion, yet one reusable CI key enrolled 181 nodes. The three controls that make that key inert.

Anthropic's Claude Mythos Preview found a 7-round AES attack in one week. Two researchers needed nearly a month to verify it. Why review is the real cost.

Clem Delangue asked OpenAI to release the rogue agent traces and commit $100M. The part that matters for defenders: hosted safety filters blocked the forensics.

Alibaba’s Claude Code ban turns a disputed backdoor warning into a practical vendor-risk checklist for AI coding teams.

One in four AI agent skills carries a security flaw. NVIDIA open-sourced SkillSpector to scan them — here's a practical pre-deploy audit playbook.

A one-line change in Claude Code 2.1.166 strips borrowed authority from relayed agent messages. Here's why the multi-agent trust boundary just moved.

Vercel deepsec shows why AI-coded apps need repeatable security harnesses, second-agent revalidation, and controlled merge gates.

OpenAI’s Codex safety post turns coding-agent adoption into a control system: sandboxing, approvals, network policy, credentials, and telemetry.

How Anthropic's new security tool finds bugs that static analysis misses — and how to use it effectively.

Week 4/2026 brings major AI updates: OpenAI tests ChatGPT ads for 800M users, Claude Cowork automates desktop tasks, Google connects Gemini to Gmail and Photos – and critical MCP security flaws reveal new risks for AI agents.

Context engineering is the discipline of curating, structuring, and defending everything that reaches the LLM at inference time. This comprehensive guide covers 2026 best practices for building reliable AI systems.