Microsoft's open source tools were hacked to steal passwords of AI developers

120 raffael_de 2026-06-09 07:42:18 UTC techcrunch.com Llama Share 42

Microsoft's Azure and AI development tools were hacked, allowing hackers to steal users' passwords and sensitive credentials. The company has temporarily removed some repositories, notified affected customers, and is continuing to investigate the breach.

Forever Young: how one molecule can lock plants in a youthful state.(2025)

16 bryanrasmussen 2026-06-09 08:42:18 UTC omnia.sas.upenn.edu Llama Share 1

Biologist Scott Poethig discovered a single molecule, miR156, that locks plants in a youthful state, a phenomenon called neoteny. This finding has big implications for conservation, horticulture, and more, as it could help scientists tailor plants to different needs and environments.

OpenCV 5 Is Here: The Biggest Leap in Years for Computer Vision

157 ternaus 2026-06-09 03:42:18 UTC opencv.org Llama Share 29

OpenCV 5 is a major release that modernizes the library with a new DNN engine, improved ONNX support, and hardware acceleration. It brings significant performance improvements, better language support, and a redesigned Hardware Acceleration Layer for faster image processing.

Apple reveals new AI architecture built around Google Gemini models

590 unclefuzzy 2026-06-08 19:42:18 UTC macrumors.com Llama Share 450

Apple overhauls Apple Intelligence with Google Foundation Models, enabling on-device AI capabilities and multimodal support. Users can opt out of Google integration, but it's unclear if this can be completely disabled.

Eagle Computer: The rise and fall of an early PC clone

4 giuliomagnifico 2026-06-09 09:20:18 UTC dfarq.homeip.net Prefix Share 0

When it comes to 80s computer brands, few flew as high as Eagle Computer flew in 1983. The aptly named company was selling 12,000 computers a month and had been doubling sales every quarter under the leadership of a talented CEO. Then Eagle lost its CEO, Dennis Barnhart, in a crashed Ferrari on the day of its IPO, June 8, 1983. In this blog post, we’ll explore the reasons Eagle Computer fell, ...

Thi.ng – open-source building blocks for computational design and art

37 nmstoker 2026-06-09 06:42:18 UTC thi.ng Llama Share 6

This site is a precursor to a semantic graph of projects, assets, and relationships, offering a tag cloud to explore 350 projects. The primary focus is on data-driven tooling for TypeScript and browser-based tech, with a broader aim at various programming approaches and languages.

Porting the ThinkPad X61 to Coreboot

70 walterbell 2026-06-09 04:42:18 UTC blog.aheymans.xyz Llama Share 22

The user successfully ported coreboot to a ThinkPad x61 using a Large Language Model (LLM) to reverse engineer the firmware, which would have taken months to accomplish manually. This achievement demonstrates the feasibility of using LLMs for reverse engineering firmware, potentially making it easier to develop open-source alternatives to closed-source firmware.

Siri AI

573 0xedb 2026-06-08 18:42:18 UTC apple.com Llama Share 531

Apple Intelligence is a helpful AI integrated into Apple devices for personal needs, protecting user privacy through on-device processing. It uses Private Cloud Compute for complex requests while maintaining user data security.

xAI is looking more like a datacentre REIT than a frontier lab

558 martinald 2026-06-08 15:42:18 UTC martinalderson.com Llama Share 437

xAI partnered with Anthropic and Google, providing them with huge capacity, and the revenue flows into SpaceX ahead of its IPO. This deal gives xAI a competitive advantage in datacentre buildout and helps alleviate compute shortages for Anthropic and Google.

Old'aVista – The most powerful guide to the old Internet

104 abnercoimbre 2026-06-09 02:42:18 UTC oldavista.com Full Share 22

Some of the wonderful people that help make Old'aVista possible:Ryan Stoddard, omf_2097, Eric, Lailing Dast, more...Click here if you'd like to contribute as well. Support Old'aVistaCommunityGuestbookThanks to For saving the internet history and making it available to all of us. This site wouldn't exist without them.

Show HN: Performative-UI – A react component library of design tropes

976 lizhang 2026-06-08 14:42:18 UTC vorpus.github.io Full Share 180

EU-banned pesticides found in rice, tea and spices

413 john-titor 2026-06-08 16:42:18 UTC foodwatch.org Llama Share 183

Pesticides banned in the EU were found in everyday food products in the Netherlands, France, Austria, and Germany. Brussels plans to weaken food safety regulations, putting consumers at risk.

MiMo-v2.5-Pro-UltraSpeed: 1T model with 1000 tokens per second

568 gainsurier 2026-06-08 15:42:18 UTC mimo.xiaomi.com Llama Share 418

Xiaomi MiMo-V2.5-Pro-UltraSpeed breaks 1000 tokens/s decode speed on a 1-trillion-parameter model. It achieves this through extreme model-system codesign with TileRT, delivering 10× the generation speed of MiMo-V2.5-Pro.

Apple Core AI Framework

298 hmokiguess 2026-06-08 19:42:18 UTC developer.apple.com Llama Share 75

Core AI helps build, run, and deploy AI models in apps using Apple silicon. It includes tools for model preparation, debugging, and performance monitoring with Xcode and the Core AI Debugger app.

Facebook is paying people overseas promoting Alberta separatism

116 vrganj 2026-06-09 06:42:18 UTC cbc.ca Llama Share 36

A Facebook investigation found that some overseas accounts, including one from Indonesia, were posing as Canadians and posting divisive content about Alberta separatism to generate income. Facebook's monetization incentives may be contributing to the spread of inauthentic content.

Looking Forward to Postgres 19: Query Hints

159 jjgreen 2026-06-08 22:42:18 UTC pgedge.com Llama Share 23

Postgres 19 introduces query advice modules pg_plan_advice and pg_stash_advice, allowing users to provide guidance to the planner without embedding hints in SQL. These modules address long-standing concerns about hints, including maintenance and upgrade headaches, and provide a more controlled and expressive way to influence query planning.

Show HN: Gitdot – A better GitHub. Open-source, written in Rust

257 baepaul 2026-06-08 19:42:18 UTC gitdot.io Full Share 225

Daily-digest podcast skill — turns a list of saved articles into a fully-produced Spotify episode using Qwen3-TTS, on top of the save-to-spotify CLI.

GoGoGrandparent (YC S16) is hiring Back end Engineers

0 2026-06-09 03:42:18 UTC ycombinator.com Llama Share

GoGoGrandparent's mission is to help older adults live independently by tailoring on-demand apps to their needs. They're hiring a Senior Backend Engineer to design scalable systems and improve the lives of seniors.

H2JVM – A Haskell Library for Writing JVM Bytecode

5 rowbin 2026-06-09 08:42:18 UTC discourse.haskell.org Prefix Share 0

Hi everyone! I have been working on a new library for writing JVM bytecode with Haskell in a nice, high level way and I’d love some feedback on it! The motivation here is for compilers to the JVM so they can focus on the actual code generation, meanwhile H2JVM takes care of all the messy details like StackMapTable analysis, label/offset resolution, etc. Here is a quick example taken from the ...

Passing DBs through continuations

56 remywang 2026-06-09 02:42:18 UTC remy.wang Llama Share 6

The author describes a shortcut to improve database performance using a technique called continuation-passing style (CPS), which allows operators to compose into tightly fused operations automatically. This approach is used in the Prela database, which maps cleanly to both a logical Entity/Relationship data model and a columnar physical storage, and achieves fast columnar execution when compiled.

Ask HN: What are tools you have made for yourself since the advent of AI?

300 aryamaan 2026-06-08 18:42:18 UTC news.ycombinator.com Llama Share 489

You're building software tools with a minimal toolchain, using a custom harness with dagger for efficiency. You're also creating a personal knowledge base with markdown search and wiki powered by Typesense for better outcomes.

FrontierCode

189 streamer45 2026-06-08 21:42:18 UTC cognition.ai Llama Share 33

FrontierCode is a benchmark that measures AI models' ability to write high-quality, maintainable code by assessing end-to-end code quality, correctness, and mergeability. Current models struggle to meet this new standard, with even the best-performing model achieving only 13.4% on the most challenging tasks.

Ask HN: Why hasn't there been a real competitor to Ticketmaster yet?

170 mdni007 2026-06-08 17:42:18 UTC news.ycombinator.com Llama Share 142

Ticketmaster maintains its market dominance through exclusive deals with venues and artists, vendor lock-in, and a business model that allows it to take a large cut of ticket sales. The company's monopoly is also perpetuated by the complicity of venues and artists who benefit from its services.

Why are cells small?

150 mailyk 2026-06-08 19:42:18 UTC burrito.bio Llama Share 68

Cells come in various shapes and sizes due to constraints on surface area-to-volume ratio and diffusion. A cell's size is determined by balancing tradeoffs between volume, diffusion, surface area, and energy.

Surveillance is not safety: A statement on the UK's latest threat to privacy [pdf]

565 g0xA52A2A 2026-06-08 20:42:18 UTC signal.org Llama Share 211

UK government's plan to scan all UK devices for nudity will not safeguard children but endanger everyone and strengthen tech giants' control. This proposal will lead to mass surveillance and censorship, expanding to target "threats" and "harmful content" beyond nudity.

How much do amd64 microarchitecture levels help in Go?

54 zdw 2026-06-09 02:42:18 UTC lemire.me Llama Share 30

The Go compiler targets a 20-year-old instruction set by default, but using a higher microarchitecture level can improve performance. Upgrading to level v2 or higher can result in significant speed gains, especially for operations like population count and set operations.

Job: Head of Stonehenge

159 mooreds 2026-06-09 03:42:18 UTC english-heritage.org.uk Llama Share 125

English Heritage is a charity that cares for over a million objects and historic sites in England, aiming to connect people to the past and each other. They are seeking a Head of Stonehenge to lead a team and deliver exceptional visitor experiences while driving financial performance.

I'm building a parallel internet, and it's called The Thinnernet

83 initramfs 2026-06-08 20:42:18 UTC inavoyage.blogspot.com Llama Share 81

The author proposes a concept called Thinnernet, which aims to provide a predictable user experience across different internet speeds by minimizing data transfer and prioritizing essential data. This idea is inspired by Steve Jobs' focus on user experience and is seen as a necessary adaptation to the internet's current limitations and the need for a more energy-efficient way to access the web.

AI is slowing down

548 crescit_eundo 2026-06-08 16:42:18 UTC wheresyoured.at Llama Share 573

The author argues that the AI bubble is unsustainable due to its massive compute commitments and revenue requirements. To justify the current infrastructure, AI companies must grow at an astonishing rate, with Anthropic and OpenAI needing to make $400 billion in annual revenue by 2030.

Apple bets cheaper AI will woo small developers

64 jbernardo95 2026-06-08 21:42:18 UTC techcrunch.com Llama Share 27

Apple is offering lower-cost AI tools to indie developers by waiving infrastructure fees for those earning under 2 million. This move reflects a growing reality in the AI industry where experimentation is no longer cheap and developers are tightening their belts.