Tech news, November 12, 2026

The 2026 APNIC Survey highlights strong confidence in APNIC’s services and the priorities of Members and stakeholders across the region.

LeRobot v0.7.0 trains on Lance tables straight from object storage, ASOF joins contributed upstream to DataFusion, and GPT-2 pretrained from one LanceDB table, plus Feature of the…

Part 1 of 2: From bare Ubuntu servers to a running MySQL cluster OpenEverest is an open source platform for running databases such as MySQL, PostgreSQL and MongoDB on Kubernetes.

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

A coordinated botnet of compromised GitHub accounts launched a widespread attack, opening seemingly harmless pull requests before pushing a malicious shell script directly to defa…

Replacement: gemini-3.8-flash.

AWS Graviton5 is now available on Amazon RDS through the new M9g instance family.

Guest Post: What happens when we replace network cables with carrier pigeons?

In October, the Internet Architecture Board (IAB) held a two-day workshop in Prague with people who are deploying post-quantum authentication.

Agent skills capture repeatable engineering work in files teams can review and reuse, while tools, project context, and permissions remain specific to the host that runs them.

JSON Web Token (JWT) validation through Azure Front Door Auth Policy is now in public preview on Azure Front Door Premium.

Mutual TLS (mTLS), also known as client certificate authentication, is now generally available on Azure Front Door Premium.

Amazon Cognito now supports OAuth 2.0 token exchange (RFC 8693).

Every year, the Google Summer of Code (GSoC) Mentor Summit brings together open source maintainers, community leads, and mentors from across the globe.

AWS says agents wrote AWS Emulator's 200+ services and security agents checked each change.

Regression testing must catch behavior changes before they reach production, yet synthetic tests miss real-world edge cases.

Reynold Xin's post "From monolith to Lakebase to LTAP" explained what LTAP is and...

What's new in research at Microsoft?

Amazon ECS Managed Instances provides built-in GPU health monitoring and automatic repair powered by NVIDIA DCGM.

GitHub Actions hosted runners will be ending support for ARM32 job execution on Windows and Linux by 8th Dec 2026.

Learn how Snowflake and its ecosystem of security innovators help enterprises govern AI agents, protect data and respond to threats across seven pillars.

See how Disney, Canva, Northern Trust, Stripe and EY are putting AI to work, from marketing ROI to finance and tax.

Token Pass now includes Kimi K3.1 Pro, Moonshot AI’s most capable model for coding and agentic workflows, with vision, tool use, and a 1M context window.

Cypress runs each test in one tab, so flows that open a new tab or popup have been hard to test.

Building multimodal AI/ML pipelines on Amazon EC2 means scaling distributed compute, controlling costs across stages, and keeping up with new models and hardware.

Learn how a distributed AI team can share packages, models, and environments across borders, with lessons from Nubank and Latin American talent.

Seedream 5.5 is an image generation and editing model from ByteDance Seed.

Gemini 4 Hydrogen is the high-efficiency workhorse model of the Gemini 4 family from Google, built for high-throughput agentic workflows, coding and multi-model routing at lower c…

Trending on the Hugging Face Hub.

New open weights from inclusionAI (Ant Group) on the Hugging Face Hub.

New open weights from inclusionAI (Ant Group) on the Hugging Face Hub.

New open weights from inclusionAI (Ant Group) on the Hugging Face Hub.

New open weights from Meta (FAIR) on the Hugging Face Hub.

Affects activestorage.

Posted by Hanno Böck on Nov 12 Hi, Forwarding this, various security issues fixed in cabextract and libmspack.

 Hello Everyone!

Bulletin ID: HCSEC-2026-57 Affected Products / Versions: Vault and Vault Enterprise up to 2.2.1; fixed in Vault 2.2.2 and Vault Enterprise 2.2.2, 1.21.14, and 1.19.25.

Bulletin ID: HCSEC-2026-56 Affected Products / Versions: Vault and Vault Enterprise up to 2.2.1; fixed in Vault 2.2.2 and Vault Enterprise 2.2.2, 1.21.14, and 1.19.25.

Bulletin ID: HCSEC-2026-55 Affected Products / Versions: Vault and Vault Enterprise up to 2.2.1; fixed in Vault 2.2.2 and Vault Enterprise 2.2.2, 1.21.14, and 1.19.25.

  The Stable channel has been updated to 158.0.8127.36/.37 for Windows, Mac will be coming shortly as part of our early stable release to a small percentage of users.

Zhenpeng Lin discovered that Apache HTTP Server's mod_dav module incorrectly calculated memory sizes when handling WebDAV locks.

Bulletin ID: 2026-182-AWS Publication Date: 11/12/2026 13:00 PM PST Description: An untrusted search path issue in Aurora PostgreSQL dialect detection in the AWS Advanced Go Wrapp…

Wolfgang Schweer discovered that Debian-LAN configured overly permissive ACLs for the Kerberos admin server.

SGLang-Neuron is now publicly available through an AWS Deep Learning Container.

Natera built SWARM, an agentic AI system on Amazon Bedrock AgentCore that captures institutional lab-operations knowledge and deploys it at scale.

The new Community Skills Feature.

Your Amazon Bedrock inference bill can keep climbing even after you tune prompts and application logic.

A zero-infrastructure  AWS DevOps Agent skill profiles your Amazon CloudWatch Logs estate and gives you one view of it.

Get per-identity Amazon Bedrock cost visibility and enforceable spend limits using only AWS-native features, with no application code for the common case.

Comparing per-token prices hides the real cost of running a model on Amazon Bedrock.

Build a self-service LLM gateway with LiteLLM on AWS Fargate that puts Amazon Bedrock foundation models behind an OpenAI-compatible API.

AI agents now run multi-step workloads for hours and act on behalf of users, raising the bar for secure, scalable agent compute.

The migration guide in the amazon-bedrock skill helps you migrate a customer-managed Amazon Bedrock knowledge base to a Managed Knowledge Base.

A secure operating system, a quantum microsystem platform, and a 5G interference-canceling algorithm are named among the year's most significant innovations.

Public sector agencies increasingly need AI spend to be a fixed, predictable number.

The generative AI Andon Cord pattern gives operators runtime controls to disable, roll back, or constrain AI features in live media workflows in seconds, without redeploying.

Speech At Yale University, New Haven, Connecticut

Every job across every industry is being reshaped by AI, but many professionals struggle to identify the right AI skills they need and where to start.

Assignar builds software for subcontractors and the general contractor firms who hire them.

Also: native mobile and frontend jobs are disappearing, how trust helps an “AI-native” engineering team to succeed, Carta starts selling a coding harness, and more

Is NVIDIA Over?

The co_await operator takes only one parameter, but it can be a big parameter.

Tomorrow I head to Washington DC for a conference called The Takeoff.

When we use LLMs in a specialized domain, such as the operations of any reasonably-large organization, there is a lot of specialized vocabulary.

Frontier video generation models work in a VAE latent space that compresses video in both space and time.

How multimodal language models should encode visual input remains up for debate.

Model architecture shapes the scaling behavior, training efficiency, and inference cost of LLMs, making it a central focus of both industry and academia.

Decoder-only Large View Synthesis Models (LVSMs) with KV-cache have recently achieved state-of-the-art quality by regressing novel views from neural networks without reconstructin…

Vision language navigation (VLN) is very challenging for Unmanned Aerial Vehicles (UAVs) since small errors in 3D motion prediction can lead to navigation errors and collisions.

Institutional portfolios operate under investment policy statements that specify asset allocation constraints, often conditioned on market regime information.

Video diffusion models are able to generate samples with high-quality appearance and textural qualities.

Applying machine learning systems to new domains introduces bottlenecks in both labeled data acquisition and manual pipeline engineering.

Given a set of LEGO parts and an image of an assembled object, our task is to infer the pose of each part to assemble the object.

Remasking allows diffusion language models to revisit generated text, but effective correction requires knowing when and what to revise.