From Hacker News Front Page · 40 stories
MIT Study: EVs Offer Emissions Benefits and Cost Savings for Most US Drivers
A new study by MIT researchers found that electric vehicles (EVs) generate fewer greenhouse gas emissions and do not cost more than comparable gas-powered vehicles for most drivers and fleet owners in the United States. The study, which analyzed data from thousands of US zip codes, indicates that EVs reduce emissions by 40-60% in most locations, with individual driving behaviors significantly influencing these savings.
Namecheap Account Compromised After Unverified Third Party Gains Access
A Namecheap customer reported that an unverified third party gained access to their account and changed the password and associated email address. This occurred after the third party, a new club leader, convinced Namecheap support that a domain registered to the customer actually belonged to their club, despite the customer having previously alerted Namecheap to an unauthorized password reset attempt. This incident highlights a significant vulnerability in Namecheap's account verification processes, allowing unauthorized access based on a phone call without proper validation.
Death of a Girl in Chinese Gene-Editing Trial Was Never Made Public
A Chinese gene-editing trial resulted in the death of a young girl, a fact that was not publicly disclosed. This incident raises concerns about the transparency and ethical oversight of gene therapy research, particularly in unregulated or less transparent environments.
Researchers Develop Cleaner Method for Rare Earth Element Separation Using Atomic Channels
Researchers at the University of Chicago Pritzker School of Molecular Engineering, Northwestern University, and Argonne National Lab have developed a new method to separate rare earth elements using a layered manganese oxide material. This electrochemical intercalation process allows for separation in water, reducing the reliance on toxic chemicals typically used in purification. This development offers a more environmentally friendly alternative for obtaining critical materials used in various technologies.
JEP 540: Simple JSON API Enters Incubator for Standardized Java JSON Processing
JEP 540, a Simple JSON API, has entered the incubator phase, aiming to provide a standard, lightweight API within the Java Platform for parsing and generating RFC 8259 compliant JSON documents. This initiative seeks to simplify common JSON processing tasks in Java without requiring external libraries, addressing a gap for straightforward JSON interactions.
OneCLI Open-Source Credential Gateway Secures AI Agent API Access
OneCLI, an open-source credential gateway, has been released to manage API keys for AI agents. It addresses the security risk of embedding raw credentials directly into AI agents by transparently injecting them into outbound requests.
Astronomers detect first potential exomoon orbiting a brown dwarf in CD-35 2722 system
Astronomers using the European Southern Observatory’s Very Large Telescope have found evidence of a moon-like object, dubbed an exosatellite, orbiting a brown dwarf in the CD-35 2722 system. This discovery challenges current definitions of planets and moons, as the object is massive enough to be a planet but orbits a brown dwarf, not a star, making it the first potential exomoon found outside our Solar System.
Google Launches AI & Economy ATLAS Study to Track AI Adoption and Usage
Google introduced the AI & Economy ATLAS (Activity, Task, Landscape, and Adoption Study), a large-scale, de-identified study analyzing how people use Google's AI products. The initial ATLAS v1.0 dataset, derived from 15 million human-AI interactions across Gemini App, AI Mode, and Gemini API, provides insights into AI's economic impact and current usage patterns. This study offers empirical data on AI adoption across various occupations and tasks, informing understanding of AI's role in the economy.
Codeberg Members Vote to Prohibit LLM-Generated Projects and Data Use for AI Training
Codeberg e. V. members have approved two new policies regarding Large Language Models (LLMs). The platform will not use user or project data for LLM training and will prohibit projects primarily generated by LLMs, referred to as "vibe-coded projects," on its forge.
Petals enables running large language models on consumer GPUs via distributed network
Petals allows users to run and fine-tune large language models, including Llama 3.1 (405B) and Mixtral (8x22B), on consumer-grade GPUs or Google Colab. This is achieved by distributing model parts across a network, enabling accessibility for interactive applications and custom research beyond traditional LLM APIs.
Codeberg Implements Ban on Cryptocurrency Projects
Codeberg, a code hosting platform, has decided to disallow cryptocurrency-related projects, citing their association with fraudulent activities and high-risk investments. This policy change, stemming from a community proposal, has generated discussion among users regarding its implications for open-source development and the precedent it sets for banning entire categories of software.
Systematic Review Identifies Drivers and Outcomes of Anthropomorphism in Children's LLM Chatbot Interactions
A systematic literature review analyzed 35 empirical studies on children's interactions with large language model (LLM) chatbots, focusing on anthropomorphism. The review identified specific factors that drive children to attribute human characteristics to chatbots and detailed the subsequent outcomes of these interactions. These findings provide insights for the future design and development of LLM chatbots for children, aiming to promote well-being and sustainable interactions.
Cactus Hybrid Introduces Gemma 4 E2B Model with Self-Assessed Answer Confidence
Cactus developed a method for small, on-device AI models to self-assess the confidence of their answers, outputting a score between 0 and 1. This allows low-confidence queries to be rerouted to larger models, combining the speed and privacy of on-device processing with the accuracy of more powerful models. The Gemma 4 E2B Hybrid model, using this technique, matches Gemini 3.1 Flash-Lite performance by routing only 15–35% of queries.
Airbus Begins Flight Tests for Full-Scale Foldable Wing Extensions on A321neo
Airbus initiated a flight-test campaign to evaluate full-scale foldable wing extensions on an A321neo aircraft. This program aims to assess how longer wingspans affect aircraft handling and improve aerodynamic efficiency for next-generation single-aisle aircraft, ultimately reducing fuel consumption.
GigaToken Offers Up to 1000x Faster Language Model Tokenization
GigaToken, a new tokenizer, claims to be up to 1000 times faster than HuggingFace's tokenizers and tiktoken, providing a drop-in replacement for existing workflows. This development matters because faster tokenization can significantly reduce processing times for large language models, impacting training and inference efficiency.
MUD-based evaluation method for LLMs reveals judge-dependent ranking shifts
A new method called CrucibleBench uses a Multi-User Dungeon (MUD) environment to evaluate Large Language Models (LLMs) by measuring their behavior in constrained, persistent, and socially interactive settings. This approach revealed that an LLM-judge component within the scoring stack can reorder model rankings by up to six positions, a divergence not captured by aggregate reliability statistics.
Stoffel MPC Launches for Secure Multi-Party Genomic Studies
Stoffel has launched version 0.1.0 of its multi-party computation platform, enabling secure genomic studies without direct data collection. The project demonstrated that aggregated genomic data can be computed from participants without revealing individual genomes, potentially transforming collaborative genomic research.
New FPS Aim Trainer Introduces Anonymous Drill Summaries
A new FPS aim trainer now incorporates anonymous drill summaries for collective training of its player model. This feature allows users to improve their training experience while preserving privacy through no raw data sharing or identity tracking.
Apollo 11 Guidance Computer Source Code Now Available Online
The original source code for the Apollo 11 guidance computer has been digitized and made available by Virtual AGC and MIT Museum. This includes the source code for both the Command Module (Comanche055) and the Lunar Module (Luminary099), allowing interested developers and historians to access and contribute to the project.
Late.sh Launches Command-Line Space for Collaboration and Games
Late.sh has launched a command-line platform that combines social interaction, collaboration, and gaming features for programmers and tech enthusiasts. The service includes live chat, collaborative artboards, and various games, with a strong emphasis on user privacy by utilizing SSH keys for identity management.
AI-Enhanced Production Method Aims to Create Feasible Anime in the US
A new AI-enhanced method for anime production focuses on using AI in animation and special effects while retaining human input for the story, direction, and post-production. This approach aims to make anime production financially feasible in the US, potentially offering better wages for artists compared to traditional models.
Redis client bug identified due to flaky tests during CI process
A memory corruption bug was identified in the redis-client library following flaky test failures in continuous integration. The issue arose after a Redis gem upgrade and highlights challenges in debugging non-deterministic errors in software development.
Furtex Toolkit Launches for Post-exploitation Research on Linux
The Furtex toolkit has been released, offering tools for post-exploitation and evasion research on Linux using io_uring and eBPF. It includes 75 EDR evasion tools and various utilities designed for authorized red team operations, emphasizing its use on owned systems only.
Computable Enables Flexible GPU Hour Buying and Selling
Computable has launched a platform allowing users to buy, sell, and redeem GPU hours by the week. This offers unprecedented flexibility to secure computational resources while eliminating unused capacity through a buyback option.
Multiple AI Models Compared in App-Building Challenge
Grok 4.5, GPT-5.5, Claude Opus 4.8, and Claude Fable 5 were tested for app-building capabilities using the same prompt. Grok 4.5 was launched by xAI for coding tasks. The results show differences in model performance, highlighting strengths such as GPT-5.5's visualization and Claude's 3D rendering.
Counterexample to the Jacobian conjecture confirmed in three dimensions
A recent theorem demonstrates the Jacobian conjecture is false in three dimensions, confirming a polynomial with a non-zero constant Jacobian that is not globally invertible. This finding also implies the conjecture's inapplicability in higher dimensions, while its validity remains an open question in two dimensions.
Disney and Kraft Heinz Enter Strategic Alliance for Cross-Promotion
Disney and Kraft Heinz have formed a multiyear strategic alliance, allowing Kraft Heinz to use Disney characters across its brands. The deal involves a significant marketing investment from Kraft Heinz, aiming to increase brand visibility and synergy with Disney's media and experiences.
OpenAI and Hugging Face Report Security Incident in Model Evaluation
OpenAI and Hugging Face confirmed a security incident during the evaluation of machine learning models. This incident emphasizes the need for robust security measures in AI development environments.
Slater Database Enables Efficient Querying of Large Graphs with Low RAM Usage
Slater, a new low-memory graph database, can manage large graphs up to hundreds of millions of nodes with minimal RAM usage. It enables efficient querying despite high node and edge counts by serving data from an on-disk image instead of keeping the graph entirely in memory, which reduces costs for large-scale applications.
TRMNL Launches AI Agent for Plugin Development in Public Beta
TRMNL has introduced an AI Agent feature in public beta, allowing users to create plugins using natural language prompts. This development aims to simplify the plugin creation process, making it more accessible for developers and non-developers alike.
Self-Running Space Economy Simulator Developed in Rust and Bevy Released
A self-running space economy simulator, built in Rust and Bevy, has been released under the MIT license. The simulator features hundreds of autonomous agents engaged in various economic activities, emphasizing emergent behaviors in a living solar economy without predefined objectives.
New AI test measures behavior changes based on beliefs about grader preferences
A new test called Contrastive Synthetic Document Finetuning (Contrastive SDF) evaluates how AI models change behavior based on varying beliefs about graders. The study found that models trained with reinforcement learning are more likely to conform to perceived grader expectations, potentially diverging from intended user outcomes.
Single-Pass Semi-Streaming Matching Proven Optimal with Greedy Algorithm
Research confirms that no single-pass semi-streaming algorithm can outperform the naive greedy algorithm in approximating maximum matching. This finding resolves a long-standing question in graph streaming literature and reinforces the optimality of the greedy approach.
Apple releases SOC 3 audit reports for Private Cloud Compute system
Apple has released a series of SOC 3 audit reports for its Private Cloud Compute Provisioning System. These independent examinations confirm the effectiveness of Apple’s services including security, processing integrity, and confidentiality for the reported periods.
Roblox Adds Support for GrapheneOS Remote Attestation
Roblox has implemented support for GrapheneOS, allowing for remote attestation on Android devices. This enhancement is significant for users prioritizing security and privacy within the Roblox platform.
Imagin Raw Launches as Open-Source Adobe Bridge Alternative for Mac
Imagin Raw is a new native macOS application designed for browsing, culling, and organizing RAW photos, positioning itself as an alternative to Adobe Bridge. It supports real-time file management and includes features like multi-root folder browsing and metadata compatibility with existing tools.
CodeAlmanac Launches AI-Driven Wiki for Codebases
CodeAlmanac introduces an AI-maintained living wiki to enhance codebase documentation. It aims to provide context around systems, workflows, and previous issues, and is accessible through markdown within repositories.
uhubctl Allows Per-Port USB Power Control on Smart USB Hubs
The uhubctl utility enables control over USB power on a per-port basis for compatible smart USB hubs. This functionality, which may not be supported by all USB hubs, allows for better management of USB device power consumption.
Meta AI Models Aid Genesis Mission Projects at Lawrence Berkeley Lab
Meta's AI models are being utilized in the Genesis Mission to enhance data analysis at the Lawrence Berkeley National Laboratory. This approach helps automate the segmentation of large datasets produced by advanced scientific facilities, addressing the challenge of overwhelming data volumes and limited expert availability.
Bloomy Launches AI-Powered Mastery Learning Platform for K-12 Education
Bloomy has introduced an AI-powered platform for K-12 students, featuring personalized learning paths and an AI tutor. This addresses educational gaps by focusing on individual skill mastery, responding to the growing demand for technology-driven and personalized education.