AI Squawk the real-time terminal for AI news

Neural surrogates for computational fluid dynamics (CFD) are trained and ranked on datasets that each vary one design family in one flow… — arXiv (AI/CL/LG, keyword-filtered)

A foundational principle of physics is that all coordinate systems are equally valid: the behavior of a PDE-governed system does not depend… — arXiv (AI/CL/LG, keyword-filtered)

Federated learning experiments can produce reasonable-looking results even when the experiment implementation differs from the intended… — arXiv (AI/CL/LG, keyword-filtered)

Learn why companies must own their agent systems, governance, context, and feedback loops to turn generic AI into lasting business… — LangChain Blog

Enforcement defenses against prompt injection, such as CaMeL, Progent, IPIGuard and DRIFT, ask the agent's own model to write a security… — arXiv (AI/CL/LG, keyword-filtered)

Embodied foundation models hinge on diverse robot interaction data, which are costly to collect. — arXiv (AI/CL/LG, keyword-filtered)

Memory decides which LLMs an edge device runs, and how fast: capacity on a $4$ GB Jetson Nano, bandwidth on an $8$ GB Jetson Orin Nano… — arXiv (AI/CL/LG, keyword-filtered)

Most business leaders I meet no longer need to be convinced that AI works. — Microsoft Cloud Blog (Azure AI)

Combining vision-language models (VLMs) with parameter-efficient fine-tuning (PEFT) offers a promising approach to few-shot whole-slide… — arXiv (AI/CL/LG, keyword-filtered)

TL;DR Helion now uses compile-time heuristics that analyze a kernel’s IR and dataflow to pick fast settings without slow autotuning. — PyTorch Blog

We've resolved the incident and the affected service is now operational. — Fireworks AI Status

TL;DR With help from AI coding agents, our two-person team integrated SGLang with IBM Spyre in less than a week, building a plugin of… — PyTorch Blog

feat: improve Codex subscription authentication guidance by @xunjizhe in #17194 … — OpenHands releases

TL;DR Scaling pretraining for a ~1-trillion-parameter Mixture-of-Experts (MoE) model on shared, public infrastructure was previously… — PyTorch Blog

Anthropic has filed to become a public company. — HN "Anthropic" (>=50pt)

All affected services are operating normally again. — OpenAI Status

Cloudflare Replicate customers may see an increase of 500 errors for predictions. — Cloudflare Status

Tip This release adds five modular model integrations for image generation, editing, video with audio, speech-driven animation, and world… — Diffusers releases (Hugging Face)

We've resolved the incident and the affected service is now operational. — Fireworks AI Status

Note text that looks like an application frame in user and tool content instead of escaping it by… — Pydantic AI releases

Added a normalization field to CartesiaTTSService.Settings and CartesiaHttpTTSService.Settings, passing Cartesia's text normalization… — Pipecat releases

A2UI 1.0 is an open protocol for generative UI that enables AI agents to render interactive, native screens across web, mobile, and desktop… — Google Developers Blog

All affected services are operating normally again. — OpenAI Status

Remove chromadb from the embedding function layer. — crewAI releases

Some customers may have experienced some elevated latency for Kimi K3.1 Pro. — Fireworks AI Status

Using an artificial intelligence agent to make flight reservations poses security risks and could violate federal rules, according to… — HN "AI" (>=75pt)

A detailed root cause analysis will be shared as soon as it is available. — GitHub Status

All affected services are operating normally again. — OpenAI Status

A detailed root cause analysis will be shared as soon as it is available. — GitHub Status

We've resolved the issue and we see recovery for all models — Baseten Status

GLM-5.3 is Z.ai's flagship model for complex software engineering and long-horizon agent tasks. — SambaNova Blog

Add the Decisions API (client.decisions.create) to the GenAI SDKs — Google GenAI JS SDK releases

It’s that time again: Introducing the world’s most powerful model, Claude Mythos 5.2. — Don't Worry About the Vase (Zvi Mowshowitz)

Ubuntu’s package servers are reachable again and template builds that install packages with apt are completing normally. — E2B Status

langchain-core 1.8.0 drops support for Python 3.10, which has reached end-of-life. — LangChain releases

We've resolved the incident and the affected service is now operational. — Fireworks AI Status

AI agents analyzed public lidar data overnight and found a Maya site under the jungle. — HN "AI" (>=75pt)

No model-provider catalog or request-admission behavior changes are included … — Open Interpreter releases

We are continuing to work on a fix for the issue. — Cloudflare Status

We've resolved the incident and the affected service is now operational. — Fireworks AI Status