Keyboard shortcuts
New here? A new edition every day at 10:00 UTC, from primary sources only — no feed, no ranking by clicks, no account. What it is · How it works →
Get the paper when it's out — each morning, and when something breaks.
Put the paper on your home screen
Like the pieces you said were worth your time. Worked out in this browser only.
Announcements from the labs, companies and platforms themselves.
The few things to know today, in two minutes →
Datasentinel 2026.10, the new release of our PostgreSQL monitoring and troubleshooting platform, is now available. It introduces AI Connector, which connects Datasentinel metrics to the AI assistant your team already uses. Ask a question in plain language, such as "why was the or…
More on Datasentinel 2026.10 · Datasentinel

The future of agents is them taking on open ended goals and driving towards completion of those goals. Learn more.

For most teams, Apache Spark™ is the better choice. Spark runs batch ETL, SQL analytics,...

Accelerate cluster access with AUP and SUP. Automated, secure user provisioning for SUNK (Slurm on Kubernetes) that syncs identities from your IdP or CW IAM in minutes.

ZGC gets adaptive heap sizing, value classes complicate numeric conversions, and primitive patterns are withdrawn as JDK 28 takes shape.
The second BETA build for the FreeBSD 15.2 release cycle is now available. ISO images for the amd64, armv7, aarch64, powerpc64, powerpc64le, and riscv64 architectures are FreeBSD mirror sites.

From major launches to real-world demos, CoreWeave’s GTC 2026 recap explores the industry shifts shaping how AI is built, run, and improved in production.

Unlock the performance of NVIDIA HGX B300 at scale on CoreWeave.

Vercel Daemons can now send OpenTelemetry traces to Vercel Tracing, allowing you and your agents to inspect background worker operations alongside the rest of your project's traces. Vercel provides each daemon with a Unix socket for receiving OTLP/HTTP traces and sets the socket'…

In June, we announced the general availability of Amazon WorkSpaces for AI agents. That launch gave your agents a managed way to operate the desktop applications your business runs on, including ERP clients, claims systems, engineering tools, and internal apps that were never bui…
More on Windows 11 · Windows

Among OpenSearch query types, span queries give you precise control over where terms appear in a document and how they relate to one another. Learn how span queries work, when to use them instead of phrase queries, and how to compose them to enforce exact term order, position, an…

An Amazon Aurora MySQL cluster can exhaust its virtual index ID limit, after which any CREATE INDEX that includes a virtual column fails with error 63955 until you rebuild the databases. This post shows how to recognize the error, why backups cannot clear it, and how to run a log…

RDS Proxy multiplexes many client connections across a small pool of database connections, but connection pinning, pool fragmentation, and uncommitted transactions can silently disable that benefit for PostgreSQL workloads. Part 1 shows how to recognize each condition with Amazon…

Metrics alone can't always pinpoint what is reducing RDS Proxy multiplexing for PostgreSQL. Part 2 shows how to use Amazon CloudWatch Logs Insights with enhanced proxy logs to find the root cause, what healthy multiplexing looks like in metrics and logs, and how to set up CloudWa…

AWS Cloud WAN Flow Logs capture flow-level traffic across your global network. Learn how to create flow logs at the core network, edge, and attachment scopes, publish them to Amazon CloudWatch Logs, Amazon S3, or Amazon Data Firehose, and query them with Amazon Athena and CloudWa…
The people building AI, writing on their own blogs. Last 30 days.

Interesting & joyful things from the previous week
This is a current list of where and when I am scheduled to speak: I’m speaking at the Catalyst Summit in Brampton, Ontario, Canada, on Wednesday, November 18, 2026. I’m speaking at QT Qonference 2026 in Toronto, Ontario, Canada, on Thursday, November 19, 2026. The list is maintai…

A decapitation. It’s time to rebuild.

November 13, 2026 — AIE speakers unite to lay out how to fight slop.
Epistemic Status: Not a physicist and obviously speculative I was pondering cosmology the other day and it struck me that the laws of physics seem surprisingly well constructed to prevent any kind of observer-accessible infinity from being observed, accessed, or utilized in any w…
Why do many people hate working in large tech companies? The obvious answer is “because organizational dysfunction stops them from doing the things they want to do”. Engineers like to write code and ship things, and if their company’s dysfunction is preventing that, it’ll make th…
Climate change is the issue. As usual, you can also use this squid post to talk about the security stories in the news that I haven’t covered. Blog moderation policy.

Why I now expect superintelligence by 2030
Tutorials, case studies and technical deep dives.






New models, as they become available.








Vulnerabilities exploited in the wild (CISA), critical advisories in the ecosystems devs build on (GitHub), and the security teams' own reports.
Posted by Osvaldo J. Filho on Nov 14 Following up on the scale-factor OOB read -- there's a second out-of-bounds read in the same decoder, in the bitstream reader. bitstream_fill_current() in bitstream.c reads a word and advances the pointer with no end-of-buffer check: stat…
Andreas Gabriel Berbescu discovered that network-manager-iodine, the iodine VPN plugin for NetworkManager, did not properly validate the nameserver setting when establishing a connection, allowing a local unprivileged user to execute arbitrary commands as root prior to privilege…
Posted by Osvaldo J. Filho on Nov 14 Hi, libdca is VideoLAN's DTS Coherent Acoustics decoder. It's effectively unmaintained (VideoLAN confirmed to me it's abandoned and it was dropped from VLC 4.0), but several distros still ship it as a library, so I'm report…
Posted by Robert Rothenberg on Nov 14 ======================================================================== CVE-2015-10149 CPAN Security Group ======================================================================== CVE ID: CVE-2015-10149 Distribution: Dancer Versions: before…
Posted by Alexander A. Shvedov on Nov 14 Product: GPAC (MP4Box) Affected: Through 26.07.0 (per the published CVE record) Reproduced on 26.08-DEV-rev263-g3f056542b-master Commit 3f056542b377cc73677d19490bfe1b46c52c9be7 Fixed: a2e5c62c213b5329a4572fe8e7c333e1ca26d028 CVE: CVE-2026-…
Posted by Alexander A. Shvedov on Nov 14 Product: GPAC (MP4Box) Affected: Through 26.07.0 (per the published CVE record) Reproduced on 26.08-DEV-rev263-g3f056542b-master Commit 3f056542b377cc73677d19490bfe1b46c52c9be7 Fixed: 43a8329a90ae2907716b593cc3746458516c7c39 CVE: CVE-2026-…
From Rust to Godot and curl, open source projects are tightening contribution rules as AI-generated code strains maintainers and changes how developers collaborate.
M-156, ChromeOS version 16836.33.0 (Browser version 156.0.8078.102)has rolled out to ChromeOS devices on the Stable channel. If you find new issues, please let us know one of the following ways: File a bug Visit our ChromeOS communities General: Chromebook Help Community Be…
Fake crypto exchange sessions are showing up on ScreenConnect servers and could be bait for technicians. See what Huntress found and what to do.
Multiple vulnerabilities have been discovered in PDNS Recursor, a resolving name server which could result in denial of service or DNSSEC validation bypass Debian trixie previously provided the 5.2 release series which is no longer supported upstream and since backports are not f…
Incidents from the status pages of the services developers build on, as each provider reports them.
| Service | Incident | Status | Reported |
|---|---|---|---|
| Bandwidth | Bandwidth Local Market Impairment - 122 - Manchester, NH | Ongoing · Critical | Nov 15, 6:10 AM UTC |
| HashiCorp | Emergency Maintenance Notification – HCP Radar | Ongoing · Minor | Nov 15, 5:26 AM UTC |
| Claude | Some users unable to sign up with Apple on the Claude iOS app | Ongoing · Major | Nov 14, 7:11 AM UTC |
| Twilio | SMS Delivery Delays from Twilio to Brisanet Brazil | Ongoing · Minor | Nov 13, 10:36 PM UTC |
| SendGrid | SMS Delivery Delays from Twilio to Brisanet Brazil | Ongoing · Minor | Nov 13, 10:36 PM UTC |
| Chargebee | Receivables Service Degradation | Ongoing · Maintenance | Nov 13, 9:07 PM UTC |
| Bitrise | Security Incident | Ongoing · Critical | Nov 13, 8:44 PM UTC |
| OpenAI | Delayed Costs Data in the Compliance API | Ongoing · Minor | Nov 13, 4:10 PM UTC |
New arXiv papers in AI, language, machine learning and software engineering.
The rapid growth of long-form documents presents a fundamental challenge to information retrieval (IR). Their length, dispersed evidence, and complex structures demand approaches that go beyond standard passage-level techniques. This survey provides a comprehensive review of long…
Retrieval-augmented generation (RAG) has attracted significant attention for enhancing large language models (LLMs) in domain-specific and knowledge-intensive tasks by utilizing external documents retrieved by retrievers. However, LLMs often struggle to determine which retrieved…
Multi-modal Recommender System (MRS) has emerged as a key information retrieval technology, widely adopted to enhance various web platforms. However, three interrelated challenges remain insufficiently explored: (1) noisy multi-modal content, (2) noisy user feedback, and (3) misa…
Recommendation throughout the cold-start period remains a challenge in recommender systems. Strictly cold entities (users/items) have no interactions, while warm-up entities only have a few unreliable interactions. Strict cold-start methods typically overlook warm-up entities, wh…
In recommender systems, user feedback typically follows a long-tail distribution, which leads many recommendation algorithms to exacerbate popularity bias by disproportionately favoring popular items. To mitigate this issue, recent studies have employed Inverse Propensity Scoring…
The conversational recommendation system (CRS) breaks the limitations of traditional static methods, presenting a novel framework for personalized recommendations with real-time adaptability and dynamic interactions. Existing methods predominantly focus on the balance between “ex…
Clusterwise Linear Regression (CLR) integrates clustering and regression to uncover complex relationships within heterogeneous datasets by simultaneously partitioning data and fitting cluster-specific models. This dual capability offers a significant advantage over traditional re…
Stable versions of the tools and libraries we track. Prereleases are left out.
| Project | Version | Date |
|---|---|---|
| llama.cpp | b12405 | Nov 15 |
| llama.cpp | b12404 | Nov 15 |
| llama.cpp | b12403 | Nov 15 |
| Neovim | Nvim development (prerelease) build | Nov 15 |
| LiteLLM | v1.108.9 | Nov 14 |
| Electron | electron v45.3.3 | Nov 14 |
| FastAPI | 0.163.0 | Nov 14 |
| Zed | collab-staging: editor: Build completion sort keys once per match (#66959) | Nov 14 |
| Google Gen AI SDK | v3.0.0 | Nov 14 |
| ClickHouse | Release v26.9.28.166-stable | Nov 14 |
New repositories under the GitHub topics we track, and packages published to npm.
Rising on GitHub this weekCoherence infrastructure for self-evolving AI agents — on the Claude Code or Codex subscription you already have.
Agent-surface implementation for devframe: installing it next to devframe enables the MCP adapter (devframe/adapters/mcp) and the devframe connect gateway.
DSH Skills Manager — 在 DeepSeek Harness 中统一加载并安全管理本机 Agent Skills · Load and safely manage local Agent Skills in DSH
Plannotator Pi extension - interactive plan review with annotations, annotate agent messages, and review code/PRs
Open-source coding-agent harness you can actually change — own the loop (prompts, gates, routing, skills, terminal states), use any model, run long tasks while you're away.
Codebase intelligence for TypeScript and JavaScript: health, complexity, duplication, architecture, styling drift, and unused code from one graph. CLI, LSP, and MCP server. Zero config for over 100 frameworks.
Local SQLite memory engine for omp agents
Markdown and HTML renderer for Svelte 5 — built for rendering streaming AI agent output from Claude Code, ChatGPT, and agentic workflows. XSS-safe defaults, streaming-aware sanitization, token caching, TypeScript types, and Svelte 5 runes.
AI at work outside tech — in health, science, finance, government and more — from the users, regulators and journals themselves. Only real use: deployments, approvals, trials, results.
Nothing in this section in the last 7 days.
Launches, keynotes, demos and talks from the makers' own channels, and interviews with the people who build it. A video loads from YouTube only when you press play.
That's the paper. The next one closes tomorrow morning.