Tech news, October 2, 2026
Friday: 557 pieces the paper read from primary sources, by section, newest first. Each links to its source; the line under it is the source's own opening.
The morning briefing
As the paper printed it on /brief/ at 22:25 UTC, in the sources' own words.
- LeadCopilot code review: API support and new default effort level
You can now request a GitHub Copilot code review through the REST and GraphQL APIs and set the review effort level for each request.
- NewsSelected models in GitHub Copilot deprecated
As of today, October 2, 2026, we have deprecated the following models across all GitHub Copilot experiences (including Copilot Chat, inline edits, ask and agent modes, and code co…
- NewsDeepSeek-Reasonix: How a poisoned config can hijack an AI coding agent
GitLab's Threat Research Group discovered a command execution vulnerability ( GHSA-grg2-7gc6-36m6, CVE-2026-102437 ) in DeepSeek-Reasonix Studio, a desktop git client designed for…
- OutageSMS Delivery Delays and Failures from Twilio to Hi3G Italy
Twilio customers may be experiencing SMS delivery delays and failures from Twilio to Hi3G network subscribers in Italy.
- SecurityTop vulnerability remediation tools in 2026
We compare the top vulnerability remediation tools in 2026: Aikido Security, Snyk, GitHub Advanced Security, Tenable, Qualys, and Wiz Category: DevSec Tools & Comparisons
- Deadlinegemini-2.5-flash-image shuts down today
Replacement: gemini-3.1-flash-image-preview.
News
97 pieces, the newest 20 here.
- Reflections on the 2026 Open Data Infrastructure Report
When my co-authors Csilla Zsigri, Adrienn Lawson, and I set out to map the state of global data infrastructure on the fifth anniversary of OpenSearch, we sought to capture more th…
- Claude Frontier Academy: $100M to train 10,000 engineers
Claude Frontier Academy trains Frontier Deployed Engineers to the standard of Anthropic’s own — a $100 million commitment to train 10,000 by the end of 2027.
- Stateless GitHub App installation tokens rolled out
The staged rollout of the stateless GitHub App installation token format, which began on April 27, 2026, is complete.
- Agentic inference optimization: 50-90% faster engines
Claude Code with Fable 5 built a Qwen 3.6 inference engine that beat vLLM by up to 90% on decode speed and 2.3x on TTFT, using MetaInfer skills on one B200.
- “They Kept Giving Me Opportunities to Grow”: Beena Emerson’s 10 Years at EDB
Software engineer Beena Emerson has been with EDB for 10 years.
- New Pathway to PyTorch Certified Associate (PTCA) Certification
We’re excited to bring you the new PyTorch Certified Associate (PTCA) Certification Pathway, combining focused learning modules and the PyTorch Certified Associate (PTCA) certific…
- Copilot code review: API support and new default effort level
You can now request a GitHub Copilot code review through the REST and GraphQL APIs and set the review effort level for each request.
- Unvalidated npm trusted publishing configurations now expire
Unvalidated npm trusted publishing configurations now expire 48 hours after creation and can no longer authorize publishing.
- npm staged publishing now supports creating new packages
You can now create a new npm package with npm stage publish, using a local session or a granular access token, including a stage-only token.
- Google Workspace Weekly Recap - October 2, 2026
Create more natural, expressive AI voiceovers in Google Vids with upgraded Gemini 3.8 Flash Lite TTS We are upgrading the underlying text-to-speech (TTS) engine powering AI voiceo…
- This Week in Open Source for October 2, 2026
by Daryl Ducharme, Open Source Programs Office This Week in Open Source for October 2, 2026 A look around the world of open source As AI becomes part of everyday software developm…
- Selected models in GitHub Copilot deprecated
As of today, October 2, 2026, we have deprecated the following models across all GitHub Copilot experiences (including Copilot Chat, inline edits, ask and agent modes, and code co…
- MedPAIR: Measuring Whether Physicians and AI Agree on What Matters in Medical QA
Yuexing Hao presents MedPAIR, a dataset comparing which sentences physicians and LLMs find relevant in clinical questions.
- 8 major updates to Cloudflare Observability
Cloudflare is launching eight major updates that bring logs, traces, analytics, alerts, dashboards, querying, and telemetry export into one observability platform, with simpler an…
- Supabase Select 2026 Recap
Build anything, operate with confidence, scale without limits.
- Scale without limits: Multigres, OrioleDB, and dbarena
The app your agent built in minutes stays on the same Postgres from prototype to petabyte.
- [In preview] Public Preview: Major version upgrades (MVU) for Azure Database for PostgreSQL elastic clusters
MVUs are now available for elastic clusters in Azure Database for PostgreSQL.
- Amazon EKS and Amazon EKS Distro now support Kubernetes version 1.37
Kubernetes version 1.37 introduced several new features and bug fixes, and AWS is excited to announce that you can now use Amazon Elastic Kubernetes Service (EKS) and Amazon EKS D…
- CoreWeave ARIA: The Coding Agent That Drives Continuous Improvement Across the AI Loop
CoreWeave ARIA is generally available in Forge.
- CoreWeave Forge: Turn AI Iteration Into Compounding Improvement
CoreWeave Forge keeps your AI loop connected and free to build on any model, framework or cloud, so every version compounds on the last.
New Models
3 pieces.
- Open model: amazon/ALoDLM-1.7B
New open weights from Amazon on the Hugging Face Hub.
- Open model: amazon/ALoDLM-8B
New open weights from Amazon on the Hugging Face Hub.
- New model: inclusionAI: Ling 3.1 Flash
Ling 3.1 Flash is a hybrid reasoning mixture-of-experts model from inclusionAI, with 25B active parameters out of 560B total.
Security Desk
57 pieces, the newest 10 here.
- Cisco Catalyst SD-WAN Manager API Authentication Bypass Vulnerability
A vulnerability in the API session-based authentication management of Cisco Catalyst SD-WAN Manager could allow an unauthenticated, remote attacker to access an affected system wi…
- Critical: gitea-runner: workflow container.options passes host namespaces and capability flags to job container when privileged mode is disabled
Affects gitea.com/gitea/runner.
- Critical: @a2ui/web_core: `openUrl` permits `javascript:` URI execution via agent-supplied button actions
Affects @a2ui/web_core.
- Critical: Vibe-Trading FastAPI endpoints permit unauthenticated access, file upload, and an RCE chain
Affects vibe-trading-ai.
- Critical: Vibe-Trading LLM-callable tools permit command execution, code injection, and SSRF
Affects vibe-trading-ai.
- Chrome Dev for Android Update
Hi everyone!
- Critical: Trigger.dev: V1 coordinator default-secret unauth Socket.IO
Affects trigger.dev.
- CVE-2026-104019 - OS command injection in the Studio Space startup script in Amazon SageMaker Distribution
Bulletin ID: 2026-125-AWS Scope: AWS Content Type: Important (requires attention) Publication Date: 10/02/2026 13:00 PM PDT Description: Amazon SageMaker Unified Studio is an AWS…
- CVE-2026-103885: Apache Directory LDAP API: Denial of service via crafted telephone number values
Posted by Emmanuel Lécharny on Oct 02 Severity: important Affected versions: - Apache Directory LDAP API 2.1.0 before 2.1.9 Description: Asymmetric Resource Consumption vulnerabil…
- CVE-2026-103878: Apache Directory LDAP API: Injection of plaintext responses during StartTLS
Posted by Emmanuel Lécharny on Oct 02 Severity: important Affected versions: - Apache Directory LDAP API 2.1.0 before 2.1.9 Description: Cleartext transmission of sensitive inform…
Technical Guides
48 pieces, the newest 10 here.
- Intent-driven development: How to guide agents from idea to implementation
- Building a High-Performance and Portable vLLM Linear Backend with Helion
TL;DR We integrated Helion into vLLM’s linear backend to explore how an autotuned, high-level kernel DSL can improve LLM inference performance while reducing kernel implementation…
- AI21 achieves an 83% reduction in time-to-start for AI workloads with AI Hypercomputer
Editor’s note: AI21 Labs is a leading global AI lab with a long track record of building foundation models, most notably the Jamba family, and today focuses on specialized LLMs an…
- Prompt caching benchmark: high cache reuse doesn’t always mean lower cost
We benchmarked prompt caching across DeepSeek, GLM, GPT, and Claude using Harbor evals and Phoenix traces, so we could compare cache reuse, estimated cost, and latency on the same…
- Deploy Oracle Database step by step on Amazon EVS with FSx for ONTAP
This post provides step-by-step procedures to deploy Oracle Database on Amazon Elastic VMware Service (Amazon EVS) with Amazon FSx for NetApp ONTAP as NFS datastore storage.
- Architect highly available Oracle Database on Amazon EVS and FSx for ONTAP
Learn how to architect a highly available Oracle Database environment on Amazon Elastic VMware Service (Amazon EVS) with Amazon FSx for NetApp ONTAP.
- Deploy open source Regional availability tools in your VPC
Deploy AWS Regional availability data as infrastructure you own.
- Cloudflare’s eBPF Replatforming Part 4: Observability, Operations, and Reliability Improvements from eBPF
The fourth instalment chronicling Cloudflare’s journey with eBPF, looking at how eBPF improved observability, incident response, and fleet security.
- How we built the fastest GLM-5 on Artificial Analysis
We achieved 200+ tokens per second on GLM-5.
- A model guide for the GPT-6 family
Learn how startups can choose GPT-6 models, tune reasoning effort, improve prompts and skills, coordinate tools, and prepare workflows for production.
AI in the Field
6 pieces.
- Computational tools for society’s most complex challenges
Associate Professor Cathy Wu uses reinforcement learning to help map out improvements to transportation and other multifaceted systems.
- Documenting the tech worker movement
Writing as a participant and researcher, PhD student JS Tan SM ’22 has co-authored a new book about the rise of tech worker protests and the employer backlash that followed.
- 3 Questions: A new resource to empower young entrepreneurs
Martin Trust Center Managing Director Bill Aulet introduces Dear Dreamer, a free platform for middle and high school students who want to learn about entrepreneurship.
- Inaugurating the Era of Super Intelligence
- Machine Learning Model Predicts Three Year Mortality in Cardiovascular Kidney Metabolic Syndrome
npj Digital Medicine, Published online: 02 October 2026; doi:10.1038/s41746-026-03261-6 Machine Learning Model Predicts Three Year Mortality in Cardiovascular Kidney Metabolic Syn…
- Clinician-centered evaluation of large language model-generated discharge summaries for longer hospitalizations
npj Digital Medicine, Published online: 02 October 2026; doi:10.1038/s41746-026-03320-y Clinician-centered evaluation of large language model-generated discharge summaries for lon…
Columns
10 pieces, the newest 6 here.
- Friday Squid Blogging: EU is Trying to Fight Unregulated Squid Fishing
The EU is recommending import controls to combat unregulated squid fishing in the Southwest Atlantic.
- 2026.40: Dots and Question Marks
The best Stratechery content from the week of September 28, 2026, including Meta's focus, what OpenAI is doing, and Mao and NBA Media Day.
- Dyson CameraJet
So when I saw Dyson had a $500 toothbrush, I was excited.
- Unidentified Flock Cameras in Florida
St.
- Inside-Out AI: Rebuilding Airbnb Behind the Scenes and Across the Guest Experience
After leading Meta’s Llama models, Ahmad Al-Dahle is now transforming Airbnb with AI — from how its teams develop products to how it serves guests.
- Why is there no Windows hot-patching support for other architectures like 32-bit ARM and MIPS?
They either predate hot-patching or never needed it.
Research
149 pieces, the newest 10 here.
- Less Decoder is More Encoder: Geometric Representation Learning from Novel View Synthesis
This paper examines the role of Novel View Synthesis (NVS) in geometric representation learning.
- MoSE3: Learning World-Space SE(3) at Every Pixel
Dense 3D point tracking has been a prominent paradigm for modeling motion in dynamic scenes, but a point track is just a 3-DoF translation curve per pixel: it captures where pixel…
- 4DCodeBench: Benchmarking Agents on Inverse Graphics of Dynamic Scenes
We introduce 4DCodeBench, a benchmark for 4D inverse graphics through code generation, in which agents reconstruct dynamic scenes from video as executable graphics programs.
- What Should World Models Forget? Stratified Retention for Continual Adaptation
Continual learning treats degradation on previously seen data as evidence of failure, a convention inherited from settings with a stationary prediction target, where a correct lab…
- RNADyn: A Benchmark for Generating and Understanding RNA Dynamics
Ribonucleic acid (RNA) functions through conformational changes that are not fully captured by static structures.
- EyeRobot 2.0: Active Gaze for Precise Manipulation without Wrist Cameras
Inspired by human vision, we introduce a framework using active gaze to enable fine-grained bimanual manipulation with only a single stereo camera.
- From Mixing to Tearing: Graph Decomposition in Decentralized Optimization via Message Passing
We study the minimization of sums of smooth strongly convex functions over undirected graphs, with each function held by one agent and communication restricted to neighbors in the…
- Unitary complexity in polynomial space
We show that if quantum commitments exist, then either there is no polynomial-time solution to the unitary synthesis problem, or $\mathsf{BPP} \neq \mathsf{NEXP}$.
- LESSER: Post-Training Data Selection with Output-Layer Gradients
The choice of post-training data for large language models substantially affects downstream performance.
- Decoding the Functional Roles of Register and High-Norm Patch Tokens in Vision Transformers
Self-supervised Vision Transformers (ViTs), such as DINOv2, learn rich visual representations, but the functions of their internal tokens remain poorly understood.
Releases
55 pieces, the newest 12 here.
- pnpm 12.9
- b11352
- MCP Python SDK 2.3.0: pip install -U mcp. Docs, https://py.sdk.modelcontextprotocol.io/
- @openrouter/sdk 1.4.21
- ONNX Runtime v1.28.3
- Desktop v0.0.43
- Claude Code 2.1.288: Ctrl+F to find a session by name and Alt+↑/↓ to jump between groups in the agents view; both, and rename…
- Temporal 1.33.0-164.0: temporalio/temporal version 1.33.0
- SGLang 0.5.21: 779 PRs from 227 contributors
- b11351
- Goose 1.53.0: Resizable Picture-in-Picture window for MCP Apps #12417
- Ollama 0.35.1: Clef decision models
New Repos & Packages
56 pieces, the newest 8 here.
Watch
16 pieces, the newest 6 here.
- Inline Charts, New Models & Jev | This Week in Replit (Oct 2, 2026)
- AWS Summit Johannesburg 2026 | AWS Events
- Azure Copilot Troubleshooting Agent is now general available
- How Pacific Life Cut Tech Debt and Sped Up AI with Snowflake
- Black Hat Stories | Tony Lee, Black Hat Review Board Member and VP of Operations at HackerOne
- Meet Qasim Sarfraz - Speaker at KubeCon + CloudNativeCon North America
Package releases
60 from the most used packages' own repositories, the newest 15 here.
- markupsafe 3.0.4 · every release →
- sqlalchemy rel_2_1_3 · every release →
- uuid v1.27.0 · every release →
- pylint v4.1.2 · every release →
- pbs-installer 2026.10.01 · every release →
- streamlit 1.65.0 · every release →
- azure-core azure-keyvault-certificates_4.11.3 · every release →
- trimesh 5.1.1 · every release →
- rou3 v0.12.0 · every release →
- pyroaring 1.2.0 · every release →
- whenever 0.11.0 · every release →
- @azure/msal-common msal-node-v7.0.1 · every release →
- eslint v10.12.0 · every release →
- libc 0.2.190 · every release →
- realtime v2.32.0 · every release →