Published on April 28, 2026 at 18:00 CEST (UTC+2)
Localsend: An open-source cross-platform alternative to AirDrop (396 points by bilsbie)
LocalSend: An open-source cross-platform alternative to AirDrop
LocalSend is a free, open-source app that enables secure file and message sharing between nearby devices over a local network without requiring an internet connection. It is designed as a cross-platform alternative to Apple’s AirDrop, emphasizing privacy and ease of use. The GitHub repository shows significant community engagement with 79.5k stars and 4.3k forks.
Microsoft VibeVoice: Open-Source Frontier Voice AI (170 points by tosh)
Microsoft VibeVoice: Open-Source Frontier Voice AI
Microsoft open-sourced VibeVoice, a unified speech-to-text model capable of processing 60-minute long-form audio in a single pass. It generates structured transcriptions with speaker identification (Who), timestamps (When), and content (What), and supports multilingual input with user-customized context. The model is now part of the Hugging Face Transformers library, making it accessible for broad integration.
Show HN: Live Sun and Moon Dashboard with NASA Footage (62 points by beeswaxpat)
Live Sun and Moon Dashboard with NASA Footage
Lumara is a real-time dashboard displaying live NASA solar imagery, moon phases, and space weather data. It uses multiple wavelengths from NASA’s SDO and SOHO satellites, plus Jean Meeus’s astronomical algorithms for offline calculations of moon rise/set and phases. The tool also tracks solar flares, CMEs, and geomagnetic storms from NASA’s DONKI database.
Deep under Antarctic ice, a long-predicted cosmic whisper breaks through (47 points by rbanffy)
Deep under Antarctic ice, a long-predicted cosmic whisper breaks through
Scientists detected a long-theorized cosmic particle (likely a neutrino) deep beneath Antarctic ice, confirming decades-old predictions. The discovery provides new insights into high-energy astrophysical phenomena and the universe’s most violent events. It validates the detection methods used by observatories like IceCube.
OpenAI CEO's Identity Verification Company Announced Fake Bruno Mars Partnership (210 points by BoggleOhYeah)
OpenAI CEO’s Identity Verification Company Announced Fake Bruno Mars Partnership
An identity verification company associated with the OpenAI CEO falsely claimed a partnership with musician Bruno Mars. The incident highlights ongoing issues with authenticity, verification, and trust in the tech industry, especially for companies dealing with identity and personal data.
Talkie: a 13B vintage language model from 1930 (535 points by jekude)
Talkie: a 13B vintage language model from 1930
Talkie is a 13-billion-parameter language model trained exclusively on text from before 1931, simulating a conversational partner from the past. Researchers use it to study how historical cultural values, knowledge, and biases are encoded in AI, and to explore “vintage” models as a tool for understanding AI behavior. The model is available for live interaction via Claude Sonnet 4.6.
UAE Leaves OPEC and OPEC+ (193 points by TechTechTech)
UAE Leaves OPEC and OPEC+
The United Arab Emirates announced its withdrawal from OPEC and OPEC+, a major shift in global oil politics. The move is expected to affect production quotas, oil prices, and geopolitical alliances in the energy market.
I Spent My Sabbatical Building a Power Meter for Sledgehammers (38 points by alin23)
I Spent My Sabbatical Building a Power Meter for Sledgehammers
A former Shopify data scientist used a paid sabbatical to create a physical power meter that measures the force and recovery impact of sledgehammer strikes. The project combines hardware prototyping, data science, and entrepreneurial testing to validate a new fitness/performance product concept.
Period tracking app has been yapping about your flow to Meta (201 points by campuscodi)
Period tracking app has been yapping about your flow to Meta
The period tracking app Flo was found liable in a class-action lawsuit for sharing sensitive user data with Meta, contradicting its privacy promises. Over 13 million users were affected, raising concerns about health data privacy and the ethics of tracking apps. The article critiques “vagueness by design” in user consent mechanisms.
GitHub Copilot code review will start consuming GitHub Actions minutes (100 points by whtsky)
GitHub Copilot code review will start consuming GitHub Actions minutes
Starting June 1, 2026, every GitHub Copilot code review on private repositories will consume GitHub Actions minutes from a user’s existing plan, with overages billed at standard rates. Usage will also be charged as AI Credits under a new usage-based model. The change reflects the operational cost of agentic AI features and shifts billing from flat-rate to consumption-based pricing.
Shift to Usage-Based Billing for AI Developer Tools
GitHub Copilot moving code reviews to a consumption model (AI Credits + Actions minutes) signals a broader industry trend: AI-assisted development features are becoming metered rather than bundled into flat subscriptions. This matters for AI/ML because it makes the cost of AI-driven workflows transparent and scalable, potentially forcing teams to optimize their usage of agentic tools. Actionably, engineering leads should audit their Copilot review volume and consider self-hosted runners to control costs.
Open-Sourcing Frontier Speech AI from Big Tech
Microsoft’s release of VibeVoice as an open-source model (integrated into Transformers) shows that major companies are investing in democratizing advanced voice AI. Key capabilities—long-form audio, speaker diarization, multilingual support—lower the barrier for startups and researchers to build voice interfaces. This trend accelerates the commoditization of speech-to-text and pushes the frontier toward real-time, context-aware conversational AI.
Vintage Language Models as a Research Tool
The “vintage” LM trained on pre-1931 text (Talkie) represents a novel methodology: using historical training data to study AI behavior, cultural bias, and knowledge evolution. This approach provides a controlled experiment for understanding how training corpus shapes model outputs. For AI/ML developers, it highlights the importance of data curation and offers a way to test alignment and safety in historical contexts.
Privacy and Trust Crises in AI-Powered Consumer Apps
The Flo period-tracking scandal—where user health data was secretly shared with Meta—underscores the tension between AI’s data-hungry nature and user privacy. As AI features become embedded in sensitive apps (health, finance, identity), regulators and users will demand stricter transparency and consent. This trend calls for privacy-by-design architectures, local-first processing (like LocalSend), and clear audit trails for data usage.
Agentic AI Raises Operational and Billing Complexity
GitHub Copilot’s agentic code review architecture relies on GitHub Actions runners, creating a new cost layer for AI features. This reflects a broader shift: agentic AI (models that call tools, run code, or interact with APIs) incurs compute and platform fees beyond the model inference cost. Teams must factor in infrastructure expenses when adopting AI agents, and platform providers will need to offer predictable billing models.
Voice AI Frontier: Unified, Long-Form, and Multimodal
VibeVoice’s ability to process one-hour audio in a single pass with structured output (speaker, timestamps, content) pushes voice AI beyond short transcriptions toward comprehensive event logging. This enables applications in meeting summarization, assistive technology, and media archiving. The trend is toward models that fuse ASR, diarization, and NLU into one pipeline—simplifying deployment but requiring careful handling of privacy for long recordings.
AI Generated Misinformation & Verification Gaps
The false partnership announcement by an identity verification company linked to the OpenAI CEO highlights a growing problem: even AI-adjacent firms can produce deceptive content, eroding trust in digital verification systems. As AI-generated content becomes indistinguishable from human-created material, the need for robust, independent verification (both technical and institutional) intensifies. This connects to the open-source movement: tools like LocalSend and vibevoice can be audited, whereas proprietary systems may hide data flows.
Analysis generated by deepseek-reasoner