Policy dependency / Stack layer
Zhipu/Z.ai Founder Tang Jie's Internal Memo 'The Giant Wave Has Arrived': Frontier AI Must Stay Open — a Direct Rebuke of China's Rumored Curbs
Bloomberg
Stack layer / Threat pattern
MCP Security by the Numbers: 30+ CVEs in a 60-Day Window, ~43% Command Injection
Practical DevSecOps
Stack layer / Threat pattern
Automated Template-Free Synthesis of Instruction-Centric Leakage Contracts for Black-Box CPUs
arXiv
Policy dependency / Stack layer
Verification is now harder than generation for coding agents — budget accordingly
arXiv 2606.26300
Stack layer / Threat pattern
Axis Intelligence AI Model Vulnerability Tracker: Agent Tool-Input Injection Succeeds 84% of the Time
Axis Intelligence Research
Stack layer / Contrast
METR: GPT-5.6 Sol Gamed Its Own Software-Engineering Eval at the Highest Rate in METR's History — Time-Horizon Score Now Unusable
METR
Stack layer / Threat pattern
GitHub Ships 'Agentic Autofix' in Public Preview — Copilot Now Explores, Fixes, and Re-Runs CodeQL Before Opening a PR
GitHub
Stack layer / Contrast
OpenScience Ships 290+ Skills and 30+ Scientific Databases as an Open Claude Science Alternative
GitHub