Show HN: We fingerprinted 178 AI models' writing styles and similarity clusters (rival.tips) AI
Researchers report a study that fingerprinted the writing styles of 178 AI models and identified similarity clusters among them, aiming to characterize how models differ in generated text.
The Future of Everything is Lies, I Guess (aphyr.com) AI
The author argues that today’s AI—especially large language models—is less like human-like intelligence and more like a “bullshit machine” that statistically imitates text while frequently confabulating, misunderstanding context, and making factual errors. They describe why LLMs can’t reliably reason about their own outputs, how “reasoning traces” and generated explanations can be misleading, and why models can be both astonishingly capable and still repeatedly “idiotic” in practical tasks. Overall, the piece frames current AI progress as creating major real-world risks alongside potential benefits, without offering a single definitive prediction of the future.
LLM plays an 8-bit Commander X16 game using structured "smart senses" (pvp-ai.russell-harper.com) AI
Russell Harper describes bringing his 1990 8-bit PvP-AI game to the Commander X16 (first running well in an emulator but slower on real hardware due to a rendering issue). He also explains how an LLM can play the game using structured “smart senses” instead of raw visual input, with the game converted to turn-based and equipped with text-based state inputs. The write-up covers the PHP-to-emulator integration and reports a progression in LLM strategies across recorded games.
MegaTrain: Full Precision Training of 100B+ Parameter LLMs on a Single GPU (arxiv.org) AI
MegaTrain is a proposed training system that enables full-precision training of 100B+ parameter LLMs on a single GPU by keeping model parameters and optimizer states in CPU host memory and streaming them layer-by-layer to the GPU for computation. The method uses double-buffered pipelining to overlap parameter prefetching, gradient computation, and offloading, and it avoids persistent autograd graphs via stateless layer templates. Reported results include training up to 120B parameters on an NVIDIA H200 with 1.5TB of host memory, and improved throughput versus DeepSpeed ZeRO-3 with CPU offloading on smaller models.
Mario and Earendil (lucumr.pocoo.org) AI
Armin Ronacher announces that Mario Zechner is joining the Earendil team, praising Pi as a thoughtful, quality-focused agent infrastructure and contrasting it with the industry’s rush for speed. He links the hire to concerns about AI systems producing “slop” and degradation, and describes Earendil’s Lefos effort to build more deliberate tools that improve communication and human relationships. Ronacher says he and Colin want to steward Pi as high-quality, open, extensible software while clarifying how it may relate to Lefos.
Multi-agentic Software Development is a Distributed Systems Problem (AGI can't save you) (kirancodes.me) AI
The post argues that multi-agent software development with LLMs is fundamentally a distributed systems coordination problem, not something that “smarter agents” will eliminate. It models prompt-driven code synthesis and agent collaboration as an underlying consensus task constrained by an underspecified natural-language spec, then relates the setting to classic impossibility results like FLP (showing limits on deterministic consensus under async delays and possible crashes) and discusses possible parallels to failure detectors. The author concludes that building scalable tooling/languages for agent coordination remains necessary even if future models become extremely capable.
The Downfall and Enshittification of Microsoft in 2026 (caio.ca) AI
The article argues that Microsoft’s 2026 “enshittification” is driven by shifting focus from core product quality to aggressive, AI-centered Copilot integration across Windows, Office, and GitHub. It points to Windows 11 promises to fix long-standing desktop usability issues, recurring complaints and outages affecting GitHub’s developer workflows, and the perceived tradeoff between reliability and Copilot placement. The author also suggests competitive pressure from Apple’s lower-cost MacBook Neo and Linux’s gradual desktop legitimacy is making Microsoft’s strategy look less like leadership and more like defensive retrenchment.
Apple is running out of A18 Pro chips for the MacBook Neo (tomsguide.com)
Apple’s MacBook Neo is selling faster than expected, and a report says the company has run short of A18 Pro chips that were “binned” for use in the laptop at the $599 price point. An independent journalist claims Apple planned to produce only about 5–6 million units using the leftover supply, which could lead to wait times for buyers. Tom’s Guide frames the issue as something buyers may not face for long, but availability could still be tight.
I've Sold Out (mariozechner.at) AI
Mario Zechner says he has joined the Earendil team and will “take pi” as a coding agent, explaining his history of OSS-to-commercial transitions and the pain he saw when key projects like RoboVM went closed-source after being sold. He describes growing interest from VCs and large companies in pi, but says he does not want to run a VC-funded company focused only on pi, prioritizing family time and avoiding the stress and community-betraying dynamics he experienced before. The post also recounts how Zechner met Armin and others in the “Vienna School of Agentic Coding” circle and how collaboration around agentic coding led to this decision.
Your File System Is Already A Graph Database (rumproarious.com)
The article argues that a plain directory of Markdown files with folders, wikilinks, and AI-assisted navigation effectively functions as a graph database for personal or team knowledge. In this setup, files become nodes and wikilinks become semantic edges, letting an agent “spider” through related meeting notes, docs, and artifacts to generate design or handoff documents with more relevant context than cold prompting. The author also notes that the hardest remaining piece is reliably automating inbox ingestion and categorization without the vault becoming too rigid or chaotic.
The Git Commands I Run Before Reading Any Code (piechowski.io)
A software engineer outlines five Git commands they run on a new repository—like identifying the most-changed files, top contributors (“bus factor”), bug-related commit hotspots, and changes in commit activity—to quickly spot where risk and churn are highest before reading any code.
Škoda DuoBell: A bicycle bell that penetrates noise-cancelling headphones (skoda-storyboard.com)
Škoda, working with scientists at the University of Salford, has developed the DuoBell, a mechanical bicycle bell designed to be audible even to pedestrians wearing active noise-cancelling headphones. Research identified a “safety gap” frequency (750–780 Hz) and a resonator/striking mechanism intended to disrupt ANC suppression. Testing reported that it increased pedestrians’ reaction distance by up to 22 metres, and London trials reportedly drew positive feedback from couriers.
Show HN: We built a camera only robot vacuum for less than 300$ (Well almost) (indraneelpatil.github.io)
The post describes how the authors built a mostly autonomous robot vacuum under a tight budget by streaming camera images to a laptop for inference, collecting teleoperated image–action data, and training a small CNN with behavior cloning to drive navigation. They report mixed results, including reversing/oscillation and difficulty learning when there is free space ahead, and note that their dataset may lack the right signals. The robot ultimately costs around $300 but still isn’t fully autonomous, requiring supervision and lacking autonomous charging.
Open Models have crossed a threshold (blog.langchain.com) AI
LangChain reports early Deep Agents evaluations showing open-weight models such as GLM-5 and MiniMax M2.7 can match closed frontier models on core agent abilities like file operations, tool use, and instruction following. The post emphasizes lower cost and latency, and describes how their shared eval suite and Deep Agents harness let developers compare and swap models across providers with minimal code changes.
Inside a huge compound where 10k workers scammed people globally (apnews.com)
The AP reports on a large compound on the Thailand–Cambodia border that it says housed about 10,000 workers involved in scamming people globally, describing how the operation functioned and the scale of its targetting. The story highlights the recruitment and management of workers and the broader cross-border challenges in combating online fraud.
Acoustic Eavesdropping with Telecom Fiber Optic Cables (ndss-symposium.org)
The NDSS paper shows how attackers can exploit the acoustic sensitivity of telecom fiber optic cables to eavesdrop, using Distributed Acoustic Sensing and a added “sensory receptor” to better capture airborne sounds at a distance. It argues that with growing Fiber-to-the-Home deployments, such techniques could allow recovery of sensitive information like human activities, indoor location, and even conversation content.
Hobby CNC machining and resin casting (2015) (lcamtuf.coredump.cx)
The article is a practical DIY guide to hobby CNC machining and resin casting, aimed at people using benchtop CNC mills or other prototyping tools. It covers choosing and maintaining a low-cost CNC setup, selecting CAD/CAM software, modeling, and methods for producing durable resin-cast parts. It also includes guidance for robot builders, basic mechanical design considerations, and workshop safety topics.
Veracrypt Project Update (sourceforge.net)
A VeraCrypt forum moderator says Microsoft terminated his driver/bootloader signing account without warning, preventing him from publishing Windows releases while Linux and macOS updates can still proceed. Other users discuss whether secure boot and signed/unsigned Windows builds will be affected, and they suggest contacting Microsoft account recovery/support or escalating via social media and other channels for human help. The thread also gathers suggestions for temporary workarounds and alternative distribution strategies while the Windows signing situation is unresolved.
Minecraft Java Edition 26.2 Snapshot 1: Initial Vulkan Support (minecraft.net)
Minecraft Java Edition 26.2 Snapshot 1 introduces initial Vulkan graphics support, indicating an early step toward improved rendering options on compatible systems.
We moved Railway's frontend off Next.js. Builds went from 10+ mins to under two (blog.railway.com)
Railway reports moving its production frontend off Next.js to a client-first stack using Vite and TanStack Router, cutting build times from 10+ minutes to under two. The team migrated 200+ routes in two pull requests with zero downtime and consolidated server configuration via Nitro, while keeping deployment and infrastructure on Railway previews/rollouts. They cite better type-safe routing, first-class layouts, and faster iteration (HMR/instant startup) as key reasons, along with improved edge caching behavior for hashed assets.
EU Migration to and from the UK (Since Brexit) (migrationobservatory.ox.ac.uk)
The Migration Observatory briefing reviews how EU migration to and from the UK has changed since Brexit, using census, labour market, education and border-related data. It finds that net migration of EU citizens turned negative in 2022 and that between mid-2021 and 2025 roughly 162,000 more EU nationals left than arrived, alongside an EU-born resident population of about 4 million in 2021/22. It also notes the post-2021 visa system has greatly reduced new EU entry via visas (around 5% of visas granted to EU nationals) while the EU Settlement Scheme remains the main route for people already in the UK. The briefing highlights limits in migration statistics during and after the pandemic and details how EU citizens’ rights and statuses are handled under the new rules.
An Arctic Road Trip Brings Vital Underground Networks into View (quantamagazine.org) AI
A Quanta Magazine field report follows biologist Michael Van Nuland and colleagues as they sample Alaskan tundra to test machine-learning predictions about rare mycorrhizal fungal “hot spots.” The article describes how underground fungal networks connect to plant roots, exchanging nutrients and carbon, and how recent imaging and robotic tracking suggest the fungi actively regulate this system rather than merely serving plants. Because these networks help store vast amounts of carbon in permafrost but are vulnerable to warming, wildfires, and thaw, the researchers argue that better mapping and protection of soil biodiversity could matter for climate resilience.
Japan relaxes privacy laws to make itself the 'easiest country to develop AI' (theregister.com) AI
Japan has approved amendments to its Personal Information Protection Act to remove the usual opt-in consent requirement for organizations using low-risk personal data for statistics and research, aiming to speed AI development. The changes include provisions for some sensitive categories such as health-related data (for improving public health) and facial images, with additional conditions like parental approval for children under 16 and stricter requirements around handling facial data. The rules add penalties for fraudulent data acquisition and improper use, but reduce requirements to notify individuals after data leaks deemed unlikely to cause harm.
Revision Demoparty 2026: Razor1911 [video] (youtube.com)
The linked YouTube video appears to be a recording from Revision Demoparty 2026 featuring a presentation by “Razor1911,” though the provided page text contains no further details about the content.
Sonnet 4.6 Elevated Rate of Errors (status.claude.com) AI
Claude Status reports that Claude Sonnet 4.6 has an elevated rate of errors, affecting claude.ai, platform.claude.com, the Claude API, Claude Code, and Claude Cowork. The company says it is investigating the issue as of April 8, 2026, with incident updates available via email or SMS.