Tech news, September 23, 2026

Radiology AI has made remarkable strides in detecting abnormalities across chest X-rays, pathology slides, and 2D scans.

Amazon Kinesis Data Streams now supports service-managed partition keys for On-Demand Standard and On-Demand Advantage streams, automatically distributing records across shards wi…

Safari Technology Preview Release 253 is now available for download for macOS Golden Gate and macOS Tahoe.

LanceDB adds ASOF join support to Apache DataFusion, improving time-series joins while giving back to the open source projects we rely on.

Reverie launches as LanceDB's summit for AI builders, GPU training bottlenecks traced to CPU-side transforms, and CrewAI unifies cognitive memory on LanceDB, plus upcoming events,…

GitHub Copilot code review now offers additional personal configurations to an expanded set of Copilot plans and an enterprise-level default setting.

This is the final notification that Node 20 is no longer available on GitHub Actions runners.

Some of the most interesting work being built with PyTorch starts in universities, research labs, student groups, and academic institutions.

Starting September 23, 2026, Microsoft is updating the author-signing certificate used for NuGet packages.

We’re expanding Google Beam to five new countries, and partnering with Industrious for an extended network.

You can now create as many Vercel Blob stores as you need.

Next.js is preparing a scheduled September security release for September 30, 2026.

AWS announces Salesforce and Zendesk data source connectors for Amazon Bedrock Managed Knowledge Base, a fully managed retrieval-augmented generation (RAG) service.

Learn what is new in Visual Studio Code 1.139 Read the full article

Support for PowerShell 7.4 will end on November 10, 2026, please transition to PowerShell 7.6.

Support for.NET 8 and.NET 9 will end on November 10, 2026.

In early results from our new life sciences research lab, Claude agents found an enzyme system whose function is still unknown.

Robots are getting smarter, but how can their hardware match that growth?

Introducing private, server-side memory to Private AI Compute for personal AI.

Ming Image 0.1 Design Layer is an image-to-image model from inclusionAI that decomposes a flattened design image into separate RGBA layers, such as a background layer and foregrou…

Gemini 3.8 Flash Lite TTS is a text-to-speech model from Google and the fast, high-throughput member of the 3.8 TTS family alongside [Gemini 3.8 Flash TTS](https://openrouter.ai/g…

Gemini 3.8 Flash TTS is a text-to-speech model from Google and the successor to [Gemini 3.1 Flash TTS Preview](https://openrouter.ai/google/gemini-3.1-flash-tts-preview).

GLM-5.3-Prime is the high-speed variant of Z.ai's GLM-5.3, inheriting its full capabilities while delivering 1.5–2× the output throughput through inference acceleration.

Qwen3.8 Max Prime is a higher-throughput variant of Qwen3.8 Max from Alibaba's Qwen team, served as a separate SKU at a higher price point.

Recraft V4.1 Flash is a text-to-image model from Recraft, the speed and cost tier of the V4.1 family.

Space Bunny Alpha is an anonymous large model with blazing-fast inference, strong coding capabilities and native multimodal input support.

Aion 3.5 Mini is a multi-model roleplaying and storytelling system from AionLabs, built on the GLM family of models.

Aion 3.5 is a multi-model roleplaying and storytelling system from AionLabs, built on the GLM family of models.

Solar Mini 4 is Upstage's compact, cost-efficient language model, a 35B-parameter mixture-of-experts with 3B active parameters and a 524K context window.

Well, well, well, well, well, well, well, well, well, well, well, well, well, well, well.

Read all about the latest AWS security features, compliance updates, and hands-on resources in our monthly digest posts.

A malicious Firefox extension fetches its payload after installation to evade detection, steal Google session cookies, and automate account takeover.

The Open Source Security Foundation (OpenSSF) is partnering with the Cloud Native Computing Foundation (CNCF) Security Technical Advisory Group (TAG Security) to support the 2026…

With Pwn2Own Ireland 2026 coming up, I wanted to share an unreleased blog post from my time as a Pwn2Own contestant.

Affects openc3.

USN-8287-1 fixed a vulnerability in XDG Desktop Portal.

Kazuma Matsumoto and Isabel Mill discovered that libgit2 incorrectly handled certain repository URLs when using the SSH transport.

Overview Imprivata Enterprise Access Management (EAM), an authentication and single sign-on platform for enterprise and clinical environments, contains a vulnerability in versions…

It was discovered that SQL parse contained multiple algorithmic complexity flaws when parsing SQL statements with deeply nested parentheses, comments, or dollar-quoted string lite…

Training a GPT-2-like model from raw text in 25 minutes with LanceDB, using one table for curation, tokenization, shuffling, packing, training, and analysis.

A GPU cluster can pass every health check and still fail to run an AI workload.

How Amazon FireTV redesigned its Continue Watching watch-progress data model on Amazon DynamoDB with vertical partitioning and hash-prefixed sort keys for parallel reads, removing…

HEMA, a 100-year-old Dutch retailer, turned developer portal-hopping into instant answers by building HAL, an internal AI assistant on Amazon Bedrock AgentCore.

An incident is the worst moment to discover you cannot trust your Amazon EKS dashboard.

How we rebuilt the diff surface in the GitHub Copilot app to open a million-line pull request with hundreds of inline review comments.

Kubernetes manages what runs on your nodes.

Learn how to build a conversational video intelligence solution on AWS using an agentic architecture.

Pair OpenCode, an open-source terminal-native AI coding agent, with open weight models on Amazon Bedrock to get a secure, flexible, pay-per-use coding assistant.

In this post, you will learn how to build a governance harness in days, and what engineering and Governance, Risk and Control (GRC) teams gain from it.

Computer scientist, entrepreneur, and philanthropist will collaborate with the MIT Schwarzman College of Computing to advance AI and scientific discovery.

Polymet is an AI product designer that turns a prompt or an image into working code.

Nature Machine Intelligence, Published online: 23 September 2026; doi:10.1038/s42256-026-01304-x Computational drug design often generates molecules that cannot be synthesized.

General methods that scale with computation will inevitably displace hand-engineered domain knowledge!

In gory detail: reliability, performance, support, pricing—and, of course, security—in our most thorough analysis of GPU cloud providers globally.

In this speech, Professor Yoshua Bengio addresses the UN Security Council to discuss the unprecedented threat posed by uncontrolled frontier AI.

Maggie Appleton on what engineers can learn from designers, working with AI agents, and why human judgment still matters.

Radical Numerics is using biological chain-of-thought and multimodal perception to keep up with the bio-defense arms race, design new genomes and gain insights into biology itself.

overshadowing more efficient GPT6 models from OpenAI

Representation Autoencoders (RAEs) enable diffusion models to operate in the feature spaces of pretrained visual encoders.

Our results show that contrastive learning outperforms a classification-based approach to authorship verification under the tested settings.

Artificial intelligence offers an unprecedented opportunity to augment human capabilities, yet progress at the frontier has focused primarily on advancing model capabilities.

Social navigation typically assumes a specified goal and focuses on reaching it while respecting social conventions, whereas robot group joining requires predicting where to join…

We introduce Sharper Transductive Local Complexity (STLC), a localized complexity method for transductive learning under uniform sampling without replacement.

Large language models (LLMs) are increasingly used in coding tasks, but their ability to reason about code execution remains unclear.

We study the nonequilibrium dynamics of a minimal recurrent transformer with $N$ normalized tokens, $Q=K=I$, and a negative value map $V=-I$.

Reordering a set of mathematical rules without changing its meaning should preserve the correct answer, but must a model's internal representations stay invariant too?

We study minimal-norm interpolation and $\ell_2$-regularized logistic-loss minimization for binary classification by univariate two-layer ReLU networks.

This work addresses continuous depression-severity score prediction from clinical interview transcripts under data scarcity.