Tech news, October 5, 2026
Monday: 526 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 00:14 UTC, in the sources' own words.
- LeadThe Agent Said It Was Done. The Database Disagreed.
- NewsOctober 03, 2026
Google SecOps SOAR Announcement Release 6.3.101 is now available for all regions.
- NewsPost-Quantum Backups Land in openSUSE Factory with ZUPT
ZUPT, an open source backup, compression, and data protection utility, has officially been accepted into openSUSE Factory, bringing a new security-focused backup option to the ope…
- OutageService Degradation Affecting Zoom Rooms Management and Workspaces Management in SA01.
The issue affecting Zoom Rooms Management and Workspaces Management has been successfully resolved.
- SecurityChromium: CVE-2025-1920 Type Confusion in V8
Information published.
- Deadlineantigravity-preview-05-2026 shuts down today
Replacement: antigravity-preview-09-2026.
News
88 pieces, the newest 20 here.
- Introducing GLM 5.3 on Amazon Bedrock
GLM 5.3 from Z.ai is now available on Amazon Bedrock: a 753B-parameter mixture-of-experts model built for coding and long-horizon agentic tasks.
- Secret scanning adds detectors for Lovable, Supabase, and more
Secret scanning now detects new secret types from Lovable Labs, Pydantic Services Inc., and Supabase.
- Let's meet in October
Scala Days 2026 is just around the corner!
- Amazon Redshift adds support for creating and refreshing Apache Iceberg materialized views
Amazon Redshift now supports the creation and refresh of Apache Iceberg materialized views.
- GLM 5.3 by Z.ai is now generally available on Amazon Bedrock
Amazon Bedrock now supports GLM 5.3 from Z.ai, giving you a powerful new option for agentic coding and long-horizon software engineering work.
- Materialize once, query anywhere: Introducing Iceberg materialized views in Amazon Redshift
Amazon Redshift now supports Iceberg materialized views.
- Railway Agent now available in Slack and Discord
Railway Agent is now available in Slack and Discord, here's how to get started and some example use cases.
- Devpack for Rust: Zero to Rust in seconds
One of the best things about Rust is how easy it is to get started.
- Open and Emergent Problems in Agentic Privacy and Security: A Contextual Angle
Generative AI
- Governed Packages and Models for Every Flow Step in Anaconda Platform
Anaconda Platform now resolves every flow step against Anaconda's governed packages and model catalog, and records what each run used.
- Evolution of the PyTorch Media Processing Landscape
TL;DR If you need to decode or encode media, whether it’s images, video, or audio, use TorchCodec.
- Use stock charts in Sheets to better visualize price movements
Google Sheets now supports new types of stock charts, making it easier to visualize price movement over time.
- MariaDB Foundation Adds PostgreSQL to Its Engine‑Agnostic Testing Framework (TAF).
The Test Automation Framework (TAF) was designed so that database makers themselves could create a plugin and integrate their database engine directly into the framework.
- Showcase your apps with new assets on the App Store
Make your apps and games stand out with new creative assets on the App Store and asset management tools in App Store Connect.
- Powering predictable costs with AWS Direct Connect flat-rate pricing
In a traditional on-premises network, predictable costs have always been central to network connectivity budgets.
- Introducing EC2 AMI Tag Sharing: Share EC2 tags across AWS accounts
EC2 Tag Sharing lets AMI owners share tags with any account that has access to the image, using the ec2:SharedTag/ prefix.
- [In preview] Public Preview: Azure HorizonDB expands to additional regions
Azure HorizonDB is expanding to additional Azure regions, giving you greater flexibility to deploy PostgreSQL workloads closer to your applications and users.With expanded regiona…
- Preview, edit and collaborate on Markdown (.md) files natively across Drive and Docs
We are introducing the ability to view, edit, and collaborate on Markdown (.md or.markdown) files directly in Google Docs, as well as view rendered previews in Google Drive.
- [Launched] Generally Available: SQL Server on Azure Virtual Machines in Azure Bleu
SQL Server on Azure Virtual Machines is now available in Azure Bleu, enabling you to deploy and manage SQL Server workloads in France's sovereign cloud environment while meeting d…
- AI Agents Need Memory: Building Persistent Memory for Agentforce
By Roopang Chauhan, Sundar Vedula, Peng-Wen Chen, and Nikhil Bojja.
New Models
2 pieces.
- New model: Tencent: Hy Image 3.5 Preview
Hy Image 3.5 Preview is a unified image generation and editing model from Tencent.
- Open model: speridlabs/iris-3b
Trending on the Hugging Face Hub.
Security Desk
51 pieces, the newest 10 here.
- USN-8872-1: libxmltok vulnerabilities
It was discovered that Expat, contained with the xmltok library did not correctly handle certain integer arithmetic.
- Critical: simple-git: `VISUAL` editor environment variable is omitted from unsafe editor detection
Affects @simple-git/argv-parser.
- Critical: simple-git unsafe-operation guard does not block trailer command configuration
Affects simple-git.
- Critical: Seroval: `fromJSON()` Promise thenable assimilation invokes plugin-produced callables (bypass of GHSA-mv8w-475r-vwqw)
Affects seroval.
- Critical: proxy-addr vulnerable to IP spoofing via IPv4-mapped IPv6 trust subnet
Affects proxy-addr.
- Critical: vm2: NodeVM custom resolution bypasses external path boundaries
Affects vm2.
- Critical: Capacitor Android and iOS: remote content can be loaded at the app origin via the internal HTTP proxy path
Affects @capacitor/android, @capacitor/ios, github.com/ionic-team/capacitor-swift-pm, com.capacitorjs:core.
- Critical: Langflow: Weak Fernet Key via random.seed()
Affects langflow.
- Critical: Langflow: Unauthenticated Flow Execution via Webhook Authentication Bypass
Affects langflow.
- AWS Continuum sets a new standard in autonomous code security
As AI models become more capable, they uncover more security vulnerabilities and identify increasingly sophisticated paths to exploit them, raising the bar for how quickly defende…
Technical Guides
41 pieces, the newest 10 here.
- Kubernetes VPC on OpenShift with Isovalent Networking for Virtualization
In this post we explore how Isovalent Networking for Virtualization with Isovalent Private Networks, and why they are best understood as the Kubernetes VPC.
- How Postman governs AI agent access to 42 production systems
How Postman gave 70 AI agents secure access to 42 production systems using Passport.
- How Snap optimizes AI storage with Amazon FSx for Lustre Intelligent-Tiering
At Snap Inc., the company behind Snapchat, cost is an important factor in every infrastructure decision.
- NEAREST BY Join: Scaling Vector Search in Databricks Runtime
Vector search originated as a serving problem.
- Connect Azure Functions to more services with managed connectors
Learn how Azure Functions managed connectors provide access to 1,700 services, with triggers and typed clients available for supported service connectors, and walk through a sampl…
- With dbt State, stop rebuilding what didn't change
Learn how dbt State lets you reuse what hasn’t changed and cut data spend.
- Protect MCP Endpoints at the Edge with Amazon CloudFront and AWS WAF
AWS WAF gives you a practical way to secure and observe Model Context Protocol (MCP) endpoints at the edge.
- Scaling Kubernetes Workloads with Node Swap
Memory is often the first hard limit a Kubernetes cluster hits.
- Supercharge regulated workloads with Claude Code and Amazon Bedrock
Anthropic Claude Opus 5.5 and Claude Sonnet 5.5 are available on Amazon Bedrock in the AWS GovCloud (US) Regions.
- UWP and WinUI 3 apps: UI testing with MSTest
Learn how to test UWP and WinUI 3 apps with MSTest and Microsoft.Testing.Platform across packaged, unpackaged, and AppContainer hosts.
AI in the Field
12 pieces, the newest 6 here.
- Guidance: AI Airlock Phase 3 application
The call for applications for phase 3 of the AI Airlock is now open.
- Policy paper: Government Response to the National Commission's Recommendations on the Regulation of AI in Healthcare.
The government's response published on the 6 October 2026 to provide clarity to stakeholders on how the government intends to take forward the recommendations made by the independ…
- 10 free Anthropic AI courses launch today on Coursera, with more coming each month to help learners stay current
Today, Anthropic, an AI safety and research company that built Claude, is launching 10 courses on Coursera that are available for free to learners around the world.
- Guidance: Good machine learning practice for medical device development
10 guiding principles that can inform the development of good machine learning practice (GMLP).
- Sovereign AI: Accelerate innovation with control and choice
Explore how Microsoft and NVIDIA are helping organizations advance sovereign AI with greater control, choice, flexibility, and resilience across cloud, edge, and customer-controll…
- Trust-Earned Autonomy: How Texas Capital Bank demonstrates an AI agent that commits to the core banking ledger, then reverses itself
Texas Capital Bank is a full-service financial services firm headquartered in Dallas, Texas, serving commercial, consumer, and institutional clients.
Columns
11 pieces, the newest 6 here.
- Anthropic Subscriptions Offer 5x+ More Value Than OpenAI
Limit testing every AI subscription plan from Anthropic, OpenAI, Meta, SpaceXSI, MiniMax, Moonshot, Z.ai, Cursor, and Cognition
- Introducing Dan Kagan-Kans
Understanding AI has a new writer!
- Why agent swarms could be the next “scaling law”
An OpenAI researcher called it “the most ‘feel the AGI’ moment that I had since reasoning models.”
- If somebody tries to hot-patch an already-hot-patched function, how do they avoid conflicts?
Hot-patching is for Windows, not for you.
- The dangerous myth behind AI agent hacks
A dangerous myth has gained traction after the recent hacks perpetrated by the AI agents developed by leading companies: That these incidents are just cyber security problems, tha…
- What on earth are you dooming about?
Paradise is not lost, the world is not ending, and most material metrics of humanity have never been better.
Research
154 pieces, the newest 10 here.
- CMU Robotics Expertise and Facilities Attract IROS 2026 Conference Attendees
Hundreds of academics and industry leaders in robotics toured Carnegie Mellon’s robotics and engineering facilities on Oct.
- Position: Stop Reactively Patching Your Model Every Time and Start Proactive Test-Driven AI Development
Position: Stop Reactively Patching Your Model Every Time and Start Proactive Test-Driven AI Development Many modern AI systems are designed to operate under diverse, open-ended, u…
- One Figure, Every Canvas: Editable Flowchart Relayout via Agentic Pipeline
Pipeline figures in ML papers must be repurposed across many canvases, including paper columns, 16:9 slides, portrait posters, 1:1 social teasers, 9:16 phone previews.
- Base Models Can Reason By Taking a Cue From Training Data
In this paper, we study how training data creates associations between the tokens at the start of a base model's response and the reasoning behavior that follows.
- InterMimicGen: Scaling Humanoid Loco-Manipulation through Self-Evolving Motion Imitation
Captured human-object interactions provide rich supervision for humanoid loco-manipulation, but they are sparse, heterogeneous, and not directly executable by robots.
- TranScope: What the Software Hides About LLM Training Data, the Hardware Reveals at Scale, and Accelerators Magnify
Membership is the root privacy primitive in machine learning: to date, no hardware-based out-of-distribution detection on black-box models has been demonstrated against constant-t…
- S2PD: Serial-to-Parallel Diffusion for Physically and Logically Consistent Video Generation
Bidirectional video diffusion models denoise entire videos in parallel, yet when trained on effectively unlimited in-distribution data from procedural generators, continue to viol…
- BiasFlow: Geometric Monitoring and Backbone Regularization for Spurious Feature Reliance
Worst-group accuracy (WGA) evaluates a trained predictor but does not characterize how its frozen backbone behaves when a new head is learned.
- Learning to Read the Contextual Tokens in Diffusion Transformers
Multimodal Diffusion Transformers (MM-DiTs) jointly process visual and textual representations throughout generation.
- Recursive Video In-Context Learning for Agentic Robot
LLM agents that orchestrate frozen vision-language-action (VLA) policies improve across episodes through text memory, which records what the agent did but not how the task is done.
Releases
40 pieces, the newest 12 here.
- Claude Code 2.1.290: ThemeKey and Color types to the plugin hooks typings, so an editor lists the theme colors a mod's drawing…
- expo-template-tabs@58.0.14
- expo-template-default@58.0.14
- @ai-sdk/workflow@2.0.60
- v0.39.0
- Nuxt 4.6.0: v4.6.0 is the next minor release
- @ai-sdk/zai@3.0.25
- @ai-sdk/xai@5.0.15
- @ai-sdk/workflow-harness@1.0.139
- 0.160.1
- ai 7.0.128
- providers-duckdb/0.2.0
New Repos & Packages
38 pieces, the newest 8 here.
Watch
11 pieces, the newest 6 here.
- GitLab Secrets Manager: Securely store secrets for pipelines, clusters, and infrastructure
- MongoDB Atlas Agent Engine | The Tradeoff Ends Here | .local NYC 2026
- How Zip Uses LangSmith and LangGraph to Ship New Features Faster
- Navigating the Future of Data Sovereignty in EMEA | Immerse Amsterdam 2026
- Secure By Default: Shipping Agents on Cloudflare | Immerse Amsterdam 2026
- Menlo’s Venky Ganesan on Going All-In on AI, Anthropic & Why Venture Has Changed Forever
Package releases
78 from the most used packages' own repositories, the newest 15 here.
- strands-agents python/v1.58.0 · every release →
- hypothesis v6.168.5 · every release →
- delta-spark v4.4.1 · every release →
- webmozart/assert 2.5.0 · every release →
- handlebars v4.7.10 · every release →
- checkov 3.3.23 · every release →
- zerocopy v0.8.60 · every release →
- azure-core azure-ai-projects_2.8.0 · every release →
- postcss-discard-duplicates cssnano@9.3.2 · every release →
- faker v40.41.0 · every release →
- nanoid 6.0.2 · every release →
- fast-copy v4.1.2 · every release →
- holidays v0.106 · every release →
- snowflake-snowpark-python v1.55.1 · every release →
- doctrine/orm 3.7.4 · every release →