Published on

Daily Tech News - 2026-05-31

Authors

The era of passive AI is officially over; today’s tech ecosystem is defined by autonomous agents, local silicon hacks, and the relentless cat-and-mouse game of cybersecurity.

Leading the daily news is a massive paradigm shift in AI development: the move from static pipelines to dynamic agents. Developers are fiercely debating the best architecture for this new world. While some argue that "backpressure is all you need" to manage unruly coding assistants, frameworks like LangGraph are rapidly becoming the standard for production data engineering. The ecosystem is maturing at breakneck speed, highlighted by Anthropic's release of Opus 4.8, the debut of the self-hosted AI workspace Odysseus, and Komi-learn bringing continuous memory to coding assistants.

To fuel these localized AI workloads, developers are getting remarkably creative. With the release of ultra-efficient models like the 1-Bit Bonsai Image 4B, the push for on-device generation is stronger than ever. Hardware hackers are even retrofitting £200 datacenter GPUs into gaming PCs just to satisfy their ever-growing VRAM cravings.

However, this rapid innovation casts a long shadow over security. Supply chain attacks remain a critical threat, as the infamous North Korean group Famous Chollima was caught targeting PHP developers through a compromised Packagist package. Furthermore, rushed AI integrations are proving perilous, with new intelligence revealing that a popular ChatGPT plugin for Google Sheets is vulnerable to data exfiltration. On the defensive side, infrastructure engineers continue to harden systems, analyzing GrapheneOS's robust server architecture and building modern data pipelines like Streambed for Postgres-to-Iceberg streaming.

Rounding out the day, the open web and media spaces saw major milestones. The VideoLAN community introduced Dav2d, a highly anticipated decoder for the emerging AV2 codec, while fed-up developers pushed for a standardized "Website Specification" to reclaim the internet from frustrating search-engine shifts.

I Put a Datacenter GPU in My Gaming PC for £200

I Put a Datacenter GPU in My Gaming PC for £200 I already had an RTX 4080. 16GB of VRAM. Good enough for gaming, not good enough for the models I wanted to run locally. The next step up in GPU land is...

  • Keywords: datacenter gpu, vram gpus, gpu vram, vram gpu, gpu adapter, gpu vm, gaming gpu, gpu slot, existing gpu, gpu tesla
  • Source: blog.tymscar.com

Backpressure is all you need

Backpressure is all you need There are two obvious ways to use coding agents. Both are bad. The first is to let the LLM run unattended and hope the repository survives. This is fast, exciting, and stu...

  • Keywords: backpressure agent, coding agents, using agent, coding agent, agentic backpressure, review agents, agent validate, agent writes, agents reviewer_agent, review agent
  • Source: lucasfcosta.com

Odysseus – self-hosted AI workspace

─────────────────────────────────────────────── ⊹ ࣪ ˖ ૮( ˶ᵔ ᵕ ᵔ˶ )っ Odysseus vers. 1.0 ─────────────────────────────────────────────── A self-hosted AI workspace -- meant to be the self-hosted version...

  • Keywords: chat_processor, agent_tools chat_processor, hosted ai, chat agents, experience chatgpt, openai agent, trojan chat, chat_processor search, chat demo, server ollama
  • Source: github.com

Restartable Sequences

May 31st, 2026 @ justine's web page The best kept secret at the frontier of system programming right now is the Linux 4.18+ (c. 2018) concept of restartable sequences or rseq for short. They allow you...

  • Keywords: rseq developed, rseq programming, restartable sequences, rseq linux, rseq supported, linux rseq, rseq memory, provides rseq, threaded programs, rseq truly
  • Source: justine.lol

Dav2d

Let dav2d be dav2d A codec does not really exist until everyone can decode it. Today, we announce dav2d, a fast decoder for the new AV2 codec, developed by members of the VideoLAN community. A few wee...

  • Keywords: dav2d codec, videolan dav2d, implementations dav2d, av2 codec, dav2d videolan, equivalents dav2d, dav2d developed, av1 decoder, directly dav2d, video codecs
  • Source: jbkempf.com

Building a LangGraph pipeline for production data engineering

LangGraph is becoming the default framework for teams building agentic AI workflows. That is both a good thing and a problem. The good part: it has real production pedigree, is actively maintained, an...

  • Keywords: workflows langgraph, build langgraph, langgraph framework, langgraph pipeline, langgraph consultant, langgraph pipelines, built langgraph, langgraph consulting, langgraph provides, does langgraph
  • Source: labyrinthanalyticsconsulting.com

1-Bit Bonsai Image 4B Image Generation for Local Devices

Introducing 1-bit and Ternary Bonsai Image 4B: Image Generation for Local Devices Today we’re releasing Bonsai Image 4B, a family of compact image-generation models designed to run high-quality diffus...

  • Keywords: image pipeline, 4b image, image generation, capable image, image model, 4b models, bonsai image, image models, image quality, 4b quality
  • Source: prismml.com

An Analysis of GrapheneOS's Server Infrastructure

An Analysis of GrapheneOS's Server Infrastructure GrapheneOS has a well-earned reputation for serious security work. Cellebrite — the forensics company law enforcement pays to crack phones — [publicly...

  • Keywords: phones grapheneos, grapheneos android, attack grapheneos, grapheneos phone, grapheneos really, grapheneos infrastructure, infrastructure grapheneos, explain grapheneos, grapheneos actually, grapheneos honest
  • Source: write.as

86Box v6.0

86Box v6.0 May 31, 2026 - written by richardg867This is the May 2026 update to 86Box, bringing in the obligatory bugfixes, some performance improvements, user interface updates and lots of new hardwar...

  • Keywords: 86box machines, using 86box, multiple 86box, support 86box, run 86box, 86box windows, update 86box, processors 86box, upgrading 86box, change 86box
  • Source: 86box.net

Build agents, not pipelines

Build agents, not pipelines There are only two ways to use LLMs in a computer program: as part of a pipeline, or as an agent. In other words, either you express the control flow of the program in code...

  • Keywords: agents pipelines, agent pipelines, pipelines agents, pipelines context, pipeline context, agent read_data_tool, pipeline agent, program pipeline, agent gather_context, read_data_tool email_tool
  • Source: seangoedecke.com

Show HN: Komi-learn – continuous memory and self-improvement for coding agents

Continuous memory and self-improvement for coding agents. It learns how you work and recalls it automatically, with no commands. Works with Claude Code and Codex. It watches a session, distills durabl...

  • Keywords: learn pip, learnings komi, komi learn, learn komi, learned komi, learns, scripts komi, recall session, agents learns, pip
  • Source: github.com

Lesson 55: Platform Engineering - Building Internal Developer Platforms

Lesson 55: Platform Engineering - Building Internal Developer Platforms What We’re Building Today Today we architect and deploy a production-grade Internal Developer Platform (IDP) on Kubernetes that...

  • Keywords: kubernetes expertise, kubernetes platform, managing kubernetes, orchestrating kubernetes, deploys kubernetes, services kubernetes, kubernetes resources, underlying kubernetes, core kubernetes, kubernetes understanding
  • Source: handsonk8s.substack.com

Famous Chollima Targets PHP Developers Through Compromised Packagist Package

Famous Chollima Targets PHP Developers Through Compromised Packagist Package The North Korean malware loader hides in a Packagist-listed package and its GitHub branch to fetch and execute remote code...

  • Keywords: malicious package, compromised packagist, developers compromised, suspicious package, javascript malware, packagist security, malicious obfuscated, malicious code, malicious javascript, branch malware
  • Source: socket.dev

How I Built a Scoreboard for My Own Agent

The bug fix took an afternoon. The follow-up question took a week. I was deep in Gemini Scribe, my Obsidian plugin that drops a Gemini-powered agent into your vault, and I had just shipped a change to...

  • Keywords: fix observability, observability gap, observability, wrote observability, bug fix, gemini scribe, observability lets, release gemini, recommending gemini, refactor bugs
  • Source: allen.hutchison.org

The Speed of Prototyping in the Age of AI

Note: These are personal reflections on how my workflow has shifted over the past year, not a pitch for any tool. Your mileage will (and probably should) vary. A few years back I wrote about my love o...

  • Keywords: meaningful workflow, thoughts ai, ai changes, think ai, warned ai, ai, ai isn, ai fits, ai magic, ai changed
  • Source: darylcecile.net

Chibil: A C compiler targeting .NET IL

Chibil is a C compiler based on chibicc rewritten in C# and updated to target .NET IL (MSIL). It is complete enough to run DOOM (PureDOOM). Chibil takes C source files and generates COFF OBJ files. Th...

  • Keywords: chibil compiler, files chibil, msvc compiler, chibil produced, chibil, net debugger, chibil probably, cli chibil, based chibicc, visual studio
  • Source: github.com

Re: [PATCH] OOM_pardon, a.k.a. don't kill my xlock

Re: [PATCH] oom_pardon, aka don't kill my xlock [Posted September 28, 2004 by corbet] On Thu, Sep 23, 2004 at 01:23:08AM +0200, Thomas Habets wrote:

How about a sysctl that does "for the love of kba...

  • Keywords: oom killed, kill processes, mechanism emergency, eject passengers, oof fuel, processes oom, oom_pardon aka, oom_pardon, victim chosen, don kill
  • Source: lwn.net

Show HN: Streambed – Stream Postgres to Iceberg on S3, Supports Postgres Wire

Postgres-to-Iceberg CDC engine. Offload analytical queries from your production database without changing your application. streambed streams WAL changes via logical replication, writes Parquet files...

  • Keywords: postgres s3, postgres iceberg, run postgres, start postgres, postgres run, postgres wal, postgres port, postgres, just postgres, query pgbench
  • Source: github.com

Avian Visitors

Avian Visitors I was initally planning on leaving this as a ‘true’ personal project of sorts. I love a good project writeup of course, but frankly I thought this was too quick an afternoon project to...

  • Keywords: avian visitors, apartment birds, bird tracking, using bird, monitor avian, user birdnet, listen birds, latest_bird home, birds avian, birdnet pi
  • Source: theodore.net

The Website Specification

What a good website does. A platform-agnostic specification of the technical features every decent website should have — from <title> to /.well-known/security.txt , from WCAG contrast to llms.txt . Wr...

  • Keywords: site security, needs seo, web platform, content standards, agents crawlers, agent readiness, seo, specification website, make site, website transport
  • Source: specification.website

Quarkus Container Image Strategy: When Buildpacks Beat Jib

Quarkus Container Image Strategy: When Buildpacks Beat Jib A practical trade-off for teams choosing between a simple app-owned image pipeline and a platform-governed build with rebase and metadata. Bo...

  • Keywords: buildpacks quarkus, quarkus buildpack, jib buildpacks, build quarkus, image buildpacks, buildpacks decision, image docker, builds image, built quarkus, image build
  • Source: the-main-thread.com

🤖 AI Weekly Recap (Week 23)

Happy Sunday! We just had another crazy week in AI. Google just dropped a 24/7 personal AI agent, while Anthropic has released Opus 4.8 and a way to orchestrate hundreds of AI agents working in parall...

  • Keywords: ai dubbing, speech dubbing, launched dubbing, dubbing v2, dubbing studio, voice clone, dubbing model, creator dubbing, io dubbing, dubbing
  • Source: simplifyingcomplexity.tech

ChatGPT for Google Sheets Exfiltrates Workbooks

Threat Intelligence Table of Content ChatGPT for Google Sheets Exfiltrates Workbooks ChatGPT for Google Sheets is vulnerable to data exfiltration and phishing overlay attacks that affect workbooks acr...

  • Keywords: workbook attacker, workbooks attack, workbooks chatgpt, sheet chatgpt, sheets vulnerable, chatgpt sheets, chatgpt excel, sheet attack, exfiltration workbooks, spreadsheets interacting
  • Source: promptarmor.com

Cloudflare Turnstile requiring fingerprintable WebGL

Cloudflare Turnstile requiring fingerprintable WebGL Since about a week, Cloudflare Turnstile (their "Verify you're human" device verification) has been looping indefinitely in my webkit-gtk based bro...

  • Keywords: disable cloudflare, fingerprintable webgl, webgl fingerprinting, cloudflare turnstile, browser fingerprinting, browser preventing, blocked webkit, webkitgtk browsers, banned webkitgtk, cloudflare device
  • Source: hacktivis.me

One &udm After Another

One &udm After Another Google made everyone mad again, so another wave of people just learned about &udm=14. Maybe we should all take the hint. “When we look back at this time, I think we will realize...

  • Keywords: demand google, google company, built google, speaking google, reason google, google founded, intrusive google, break google, turn google, ai google
  • Source: feed.tedium.co

Getting Started with the Pi Agent Kit on Docker Sandboxes

Run the Pi coding agent in an isolated microVM in about ten minutes, with your API key kept on the host and out of the agent's reach. A step-by-step walk through the official Pi kit for Docker Sandbox...

  • Keywords: docker sandboxes, docker, isolated microvm, kit docker, microvm, microvm minutes, sandboxes, host agent, run pi, api key
  • Source: ajeetraina.com

Please don't mess with links: (Rants) (Web sites)A link is just a button that takes you somewhere when you click it right? <style >span {text-decoration :underline ;color :#44EEEE ; } </style > <span...

  • Keywords: link span, links javascript, open links, link loading, buttons links, links loading, span onclick, onclick window, link previews, mess links
  • Source: maurycyz.com