A study found that AI chatbots provide differing financial recommendations across platforms and may be biased depending on demographic groups.
A study has found that text-based guardrails fail to properly defend against attacks carried out through tool-call sequences in LLM agents.
IBM and Red Hat have launched Lightwell to defend against attacks exploiting AI-discovered security vulnerabilities.
LingBot-Video, an action-conditioned world model that predicts robot motion using a Sparse-MoE architecture, has been released.
SpaceXAI has released Grok 4.5, its latest model optimized for coding, agentic tasks, and knowledge work.
VS Code 1.128 strengthens BYOK support and updates the Agents window feature.
NVIDIA has released the Nemotron open datasets and a visualization tool to improve agent performance.
Cognition unveiled SWE-1.7, a frontier-level intelligence model that maximizes cost efficiency through improvements to its RL pipeline.
The first lawsuit has been filed against OpenAI claiming that an AI chatbot's responses induced a violent crime.
Flint, an open-source visualization language that helps AI agents generate expressive charts, has been released.
Using GitHub Copilot CLI and the Namecheap API, you can connect a custom domain to GitHub Pages in just 14 minutes without going through complicated DNS setup.
LangChain and NVIDIA have launched the NemoClaw blueprint for building high-performance agents for enterprises.
NVIDIA NemoClaw enables running Deep Agents code in a security-hardened environment.
Claude Code's system prompt has been updated for agent behavior, data schema, and guide routing.
llama.cpp's ggml library has added Q2_0 quantization for CPU to support Ternary Bonsai models.
Chinese authorities warned that a security backdoor with data leakage risks was discovered in Anthropic's coding agent Claude Code.
Anthropic has unified Claude's web and desktop environments and introduced remote session and scheduled task capabilities.
Mistral AI has unveiled Robostral Navigate, its first model for robots to navigate physical space.
OpenAI has announced principles for government and national security partnerships focused on responsible AI use and public safety.
OpenAI announced that about 30% of the SWE-bench Pro benchmark contains errors due to design flaws.
A UN-affiliated scientific panel announced that current regulatory and governance tools are inadequate compared to the pace of AI technology advancement.
AI21 proposes a strategy that optimizes cost and speed for SWE agents by flexibly allocating compute resources according to task difficulty.
OpenAI is partnering with the Walton Family Foundation to host the 'AI Skills Jam,' aimed at strengthening practical AI skills among K-12 educators.
It's an AI agent tool that combines a map visualization feedback loop for geospatial data analysis.
AI Briefing