Security

Beyond GRPO and On-Policy Distillation: An Empirical Sparse-to-Dense Reward Principle for Language-Model Post-Training

Abstract Training efficiency is improved by strategically allocating scarce labeled data through staged reinforcement learning and dense supervision, using sparse rewards for teacher model discovery and dense rewards for student model compression. AI-generated summary In…

Latest Version of Anthropic’s Mythos AI is Even Better at Hacking, UK Researchers Say

Anthropic’s latest version of its Mythos AI showed “notable capability jumps” at finding and exploiting undiscovered software vulnerabilities compared to an earlier version of the model, researchers at the U.K.’s AI Security Institute said Wednesday. Anthropic has not released…

Open Source Resistance: keep OSS alive on company time

Article URL: https://ossresistance.com/ Comments URL: https://news.ycombinator.com/item?id=48123015 Points: 215 # Comments: 70

TextGen is now a native desktop app. Open-source alternative to LM Studio (formerly text-generation-webui).

Hi all, I have been making a lot of updates to my project, and I wanted to share them here. TextGen (previously text-generation-webui, also known as my username oobabooga or ooba) has been in development since December 2022, before LLaMa and llama.cpp existed. In the last two…

Large Language Models over Networks: Collaborative Intelligence under Resource Constraints

Abstract Collaborative intelligence enables multiple distributed LLMs to work together across devices and clouds to provide high-quality responses under diverse resource constraints. AI-generated summary Large language models (LLMs) are transforming society, powering…

The Trillion-Parameter Dilemma: MiMo-V2.5-Pro went open-source (1.02T params). Is self-hosting worth it when the API costs $70 for 387M tokens?

Xiaomi open-sourced MiMo-V2.5-Pro. 1.02 trillion parameters, 42B active (MoE), 1M context, MIT license. On paper, this is exciting. In practice, I'm stuck on the math. What I've been doing with it I've been running V2.5-Pro via the API through Claude Code for autonomous coding…

HEPA: A Self-Supervised Horizon-Conditioned Event Predictive Architecture for Time Series

arXiv:2605.11130v1 Announce Type: new Abstract: Critical events in multivariate time series, from turbine failures to cardiac arrhythmias, demand accurate prediction, yet labeled data is scarce because such events are rare and costly to annotate. We introduce HEPA…

DeconDTN-Toolkit: A Library for Evaluation and Enhancement of Robustness to Provenance Shift

arXiv:2605.11237v1 Announce Type: new Abstract: Despite the burgeoning body of work on distribution shifts, provenance shift-where the relationship between data source and label changes at deployment-remains poorly understood and under-addressed. In this paper, we establish a…

Towards Visually-Guided Movie Subtitle Translation for Indic Languages

arXiv:2605.11993v1 Announce Type: new Abstract: Movie subtitle translation is inherently multimodal, yet text-only systems often miss visual cues needed to convey emotion, action, and social nuance, especially for low-resource Indic languages (English to Hindi, Bengali, Telugu,…

Sign Language Recognition and Translation for Low-Resource Languages: Challenges and Pathways Forward

arXiv:2605.12096v1 Announce Type: new Abstract: Sign languages are natural, visual-gestural languages used by Deaf communities worldwide. Over 300 distinct sign languages remain severely low-resource due to limited documentation, sparse datasets, and insufficient computational…

AgentShield: Deception-based Compromise Detection for Tool-using LLM Agents

arXiv:2605.11026v1 Announce Type: cross Abstract: Defenses against indirect prompt injection (IPI) in tool-using LLM agents share two structural weaknesses. First, they all attempt to prevent attacks rather than detect the compromises that slip through. Second, they have only…

Fine-Tuning TranslateGemma-4B to improve bi-directional English & Welsh translations on an H200 GPU!

Open source repo: https://github.com/grctest/finetuned-gemmatranslate-cy 5% of the fine-tuning took 40 minutes and cost a couple dollars to prove the process works. Looking forwards to Flash Attention v4 to leave beta, to test fine-tuning performance on a B200 on the cloud,…

CERT is releasing six CVEs for serious security vulnerabilities in dnsmasq

Article URL: https://lists.thekelleys.org.uk/pipermail/dnsmasq-discuss/2026q2/018471.html Comments URL: https://news.ycombinator.com/item?id=48112042 Points: 206 # Comments: 96

Bambu Lab is abusing the open source social contract

Article URL: https://www.jeffgeerling.com/blog/2026/bambu-lab-abusing-open-source-social-contract/ Comments URL: https://news.ycombinator.com/item?id=48109224 Points: 1041 # Comments: 351

Thoughts on GitLab's workforce reduction" and "structural and strategic decisions"

GitLab Act 2 There's a lot going on in this announcement from GitLab about the "workforce reduction" and "structural and strategic decisions" they are making with respect to the agentic era. They're "planning to reduce the number of countries by up to 30% where we have small…

[AINews] Anthropic growing 10x/year while everyone else is laying off >10% of their workforce

A quiet day lets us reflect on an interesting dichotomy in the economy.

Model Quantization: Post-Training Quantization Using NVIDIA Model Optimizer

Model quantization is an effective method to reduce VRAM usage and improve inference performance on consumer devices such as NVIDIA GeForce RTX GPUs. By...

langchain-core==0.3.86

Changes since langchain-core==0.3.85 release(core): 0.3.86 ( #37242 ) fix(core): backport path-traversal fix to v0.3 ( CVE-2026-34070 , GHSA-qh6h-p6c9-ff54 ) ( #37233 )

Scaling Trusted Access for Cyber with GPT-5.5 and GPT-5.5-Cyber

OpenAI expands Trusted Access for Cyber with GPT-5.5 and GPT-5.5-Cyber, helping verified defenders accelerate vulnerability research and protect critical infrastructure.

Secure Marketplace credentials with Production-only access

You can now secure native integration resources by restricting where they can be used. Setting a resource to Production only removes non-production access and protects credentials as sensitive environment variables . This makes it so secret values or no longer readable from the…

How KIKO Milano scales for Black Friday

KIKO Milano on Vercel: Eliminated 3 weeks of Black Friday infrastructure prep 75% decrease in app build times Went from minimal releases to deploying multiple times per day KIKO Milano’s ecommerce team used to treat peak traffic as an operations project. Weeks before Black…

Introducing deepsec: The security harness for finding vulnerabilities in your codebase

Today we’re open sourcing deepsec : a security harness powered by coding agents. It runs on your own infrastructure and surfaces hard-to-find issues in large codebases. You can run deepsec on your laptop without setting up a cloud service for privileged source code access. For…

An open-source spec for orchestration: Symphony

Learn how Symphony, an open-source spec for Codex orchestration, turns issue trackers into always-on agent systems—boosting engineering output and reducing context switching.

Lights, camera, open source!

Ryan is joined on the show by Cult.Repo producers Emma Tracey and Josiah McGarvie to discuss making documentaries about open-source software and the people behind the major technologies that uphold the internet.

Maximizing Memory Efficiency to Run Bigger Models on NVIDIA Jetson

The boom in open source generative AI models is pushing beyond data centers into machines operating in the physical world. Developers are eager to deploy these...

Bringing AI-driven protein-design tools to biologists everywhere

Founded by Tristan Bepler PhD ’20 and former MIT professor Tim Lu PhD ’07, OpenProtein.AI offers researchers open-source models and other tools for protein engineering.

Ecom-RLVE: Adaptive Verifiable Environments for E-Commerce Conversational Agents

not much happened today

**OpenAI** expanded its Agents SDK by separating the agent harness from compute/storage, enabling long-running, durable agents with features like file/computer use, skills, memory, and compaction. The harness is now open-source and supports execution via partner sandboxes,…

Hack the AI agent: Build agentic AI security skills with the GitHub Secure Code Game

Learn to find and exploit real-world agentic AI vulnerabilities through five progressive challenges in this free, open source game that over 10,000 developers have already used to sharpen their security skills. The post…

Running Large-Scale GPU Workloads on Kubernetes with Slurm

Slurm is an open source cluster management and job scheduling system for Linux. It manages job scheduling for over 65% of TOP500 systems. Most organizations...

Summary of CVE-2026-23869

Summary A high-severity vulnerability (CVSS 7.5) in React Server Components can lead to Denial of Service. We created new rules to address these vulnerabilities and deployed them to the Vercel WAF to automatically protect all projects hosted on Vercel at no cost. However, do not…

not much happened today

**Google** introduced **Skills in Chrome**, enabling reusable browser workflows with Gemini prompts and a library of ready-made Skills, enhancing end-user agentization. **Tencent** teased **HYWorld 2.0**, an open-source 3D world model generating editable scenes from a single…

Anthropic Accidentally Leaked the Secret Roadmap of Claude Code

The source code of Claude Code reveals unreleased features, internal codenames, and the future of your new favorite AI product. Here's what it all means.

Training mRNA Language Models Across 25 Species for $165

Back to Articles Training mRNA Language Models Across 25 Species for $165 Team Article Published March 31, 2026 Upvote 27 Maziyar Panahi MaziyarPanahi OpenMed Part II: Building the Pipeline, From Structure Prediction to Codon Optimization By OpenMed, Open-Source Agentic AI for…

Transfer Marketplace resources between teams

You can now transfer Marketplace resources between teams directly from the Vercel dashboard without relying on the API. This simplifies resource management during team or project changes. Both owner and member roles on the source and destination teams can initiate transfers. The…

Making Turborepo 96% faster with agents, sandboxes, and humans

Turborepo is now 81-91% faster to compute its task graph in our repositories, scaling with repo size. On our 1,000+ package monorepo, turbo run now feels instant. Time to First Task is now 11x faster. After testing my changes with some open source Turborepos and asking Vercel…

The Claude Code Source Leak

**Anthropic's** closed-source coding product **Claude Code** experienced a significant source leak exposing over **500k lines** of orchestration logic, including autonomous modes and memory systems, but not model weights. The leak led to rapid public reverse-engineering,…

not much happened today

**Cursor's Composer 2**, built on **Kimi K2.5**, sparked discussion over model attribution and licensing, highlighting a shift toward post-trained derivatives of open-source models with domain-specific fine-tuning and reinforcement learning. **Claude Code** is expanding into…

AI is becoming a second brain at the expense of your first one

The risk isn’t just that we’ll get lazy and become lousy at critical thinking; the risk is that we’ll outsource our judgement and lose the ability to make qualitative, moral, and interpersonal judgments altogether.

State of Open Source on Hugging Face: Spring 2026

Back to Articles State of Open Source on Hugging Face: Spring 2026 Team Article Published March 17, 2026 Upvote 86 Avijit Ghosh evijit huggingface Lucie-Aimée Kaffee frimelle huggingface Yacine Jernite yjernite huggingface Irene Solaiman irenesolaiman huggingface This post…

Keeping the lights on for open source

Ryan sits down with Chainguard CEO Dan Lorenc to chat about how his team is keeping the foundation of the internet—open source projects—alive by forking archived but widely-used repos to provide security maintenance and dependency upgrades.

Vercel Open Source Program: Winter 2026 cohort

In April, we launched the Vercel Open Source Program to give maintainers the resources, credits, and support they need to ship faster and scale confidently. The first group joined through our spring 2025 cohort . Today we are welcoming the winter 2026 cohort. From AI-native apps…

🎂 ThursdAI — 3rd BirthdAI: Singularity Updates Begin with Auto Researcher, Uploaded Brains, OpenClaw Mania & NVIDIA's $26B Bet on Open Source

From Weights & Biases, celebrate our 3rd year in a row covering AI news, with Karpathy's mini singularity, China's full embrace of OpenClaw and uploaded brains + 3 interviews this week, don't miss!

Keep the Tokens Flowing: Lessons from 16 Open-Source RL Libraries

Back to Articles Keep the Tokens Flowing: Lessons from 16 Open-Source RL Libraries Published March 10, 2026 Update on GitHub Upvote 151 Amine Dirhoussi aminediroHF Quentin Gallouédec qgallouedec Kashif Rasul kashif Lewis Tunstall lewtun Edward Beeching edbeeching Albert…

Olmo Hybrid and future LLM architectures

The latest Olmo model and discussions at the frontier of open-source post training tools.

Develop Native Multimodal Agents with Qwen3.5 VLM Using NVIDIA GPU-Accelerated Endpoints

Alibaba has introduced the new open source Qwen3.5 series built for native multimodal agents. The first model in this series is a ~400B parameter native...

Maximizing GPU Utilization with NVIDIA Run:ai and NVIDIA NIM

Organizations deploying LLMs are challenged by inference workloads with different resource requirements. A small embedding model might use only a few gigabytes...

Unlock Massive Token Throughput with GPU Fractioning in NVIDIA Run:ai

As AI workloads scale, achieving high throughput, efficient resource usage, and predictable latency becomes essential. NVIDIA Run:ai addresses these challenges...

📆 Open source just pulled up to Opus 4.6 — at 1/20th the price

Plus: Gemini 3 Deep Think hits 84% on ARC-AGI, OpenAI's new 1000 t/s coding model, and the video model that shattered reality.

The Future of the Global Open-Source AI Ecosystem: From DeepSeek to AI+

Back to Articles The Future of the Global Open-Source AI Ecosystem: From DeepSeek to AI+ Team Article Published February 3, 2026 Upvote 53 Adina Yakefu AdinaY huggingface Irene Solaiman irenesolaiman huggingface This is the third and final blog in a three-part series on China's…

Introducing Daggr: Chain apps programmatically, inspect visually

Back to Articles Introducing Daggr: Chain apps programmatically, inspect visually Published January 29, 2026 Update on GitHub Upvote 107 merve merve yuvraj sharma ysharma Abubakar Abid abidlabs hysts hysts Pedro Cuenca pcuenq TL;DR: Daggr is a new, open-source Python library for…

not much happened today

**AI News for 1/27/2026-1/28/2026** highlights a quiet day with deep dives into frontier model "personality split" where **GPT-5.2** excels at *exploration* and **Claude Opus 4.5** at *exploitation*, suggesting **OpenAI** suits research workflows and **Anthropic** commercial…

Architectural Choices in China's Open-Source AI Ecosystem: Building Beyond DeepSeek

Back to Articles Architectural Choices in China's Open-Source AI Ecosystem: Building Beyond DeepSeek Team Article Published January 27, 2026 Upvote 45 Adina Yakefu AdinaY huggingface Irene Solaiman irenesolaiman huggingface This is the second blog in a three-part series on…

One Year Since the “DeepSeek Moment”

Back to Articles One Year Since the “DeepSeek Moment” Team Article Published January 20, 2026 Upvote 62 Adina Yakefu AdinaY huggingface Irene Solaiman irenesolaiman huggingface This is the first blog in a series that will examine China’s open source community’s historical…

Salesforce rolls out new Slackbot AI agent as it battles Microsoft and Google in workplace AI

Salesforce on Tuesday launched an entirely rebuilt version of Slackbot , the company's workplace assistant, transforming it from a simple notification tool into what executives describe as a fully powered AI agent capable of searching enterprise data, drafting documents,…

Nous Research's NousCoder-14B is an open-source coding model landing right in the Claude Code moment

Nous Research , the open-source artificial intelligence startup backed by crypto venture firm Paradigm , released a new competitive programming model on Monday that it says matches or exceeds several larger proprietary systems — trained in just four days using 48 of…

not much happened today

**Alibaba** released **Qwen-Image-Layered**, an open-source model enabling Photoshop-grade layered image decomposition with recursive infinite layers and prompt-controlled structure. **Kling 2.6** introduced advanced motion control for image-to-video workflows, supported by a…

Codex is Open Sourcing AI models

Back to Articles Codex is Open Sourcing AI models Published December 11, 2025 Update on GitHub Upvote 82 ben burtenshaw burtenshaw shaun smith evalstate Building on our work to get Claude Code to train open source models, we are now getting Codex to go further. We gave Codex…

We Got Claude to Fine-Tune an Open Source LLM

Back to Articles We Got Claude to Fine-Tune an Open Source LLM Published December 4, 2025 Update on GitHub Upvote 624 ben burtenshaw burtenshaw shaun smith evalstate We gave Claude the ability to fine-tune language models using a new tool called Hugging Face Skills . Not just…

Hired Through GitHub: Part 2

More stories from the open source contributors who became core team members.

Rethinking how we measure AI intelligence

Game Arena is a new, open-source platform for rigorous evaluation of AI models. It allows for head-to-head comparison of frontier systems in environments with clear winning conditions.

Hired Through GitHub: Part 1

Stories from the open source contributors who became core team members.

Understanding and Implementing Qwen3 From Scratch

A Detailed Look at One of the Leading Open-Source LLMs

Beating ARC the hard way

ARC is benchmark developed to test out of distribution reasoning and common sense in general solvers. It is specifically designed to be: Easily solvable by most humans Not amenable to any kind of brute-force solvers (e.g. try every permutation of a solution) Not able to be…

Reward Hacking in Reinforcement Learning

Reward hacking occurs when a reinforcement learning (RL) agent exploits flaws or ambiguities in the reward function to achieve high rewards, without genuinely learning or completing the intended task. Reward hacking exists because RL environments are often imperfect, and it is…

What I learned from looking at 900 most popular open source AI tools

[ Hacker News discussion , LinkedIn discussion , Twitter thread ] Update (Feb 2026) : The full list of open source AI repos is hosted at Good AI List , updated daily. It’s balooned to 15K repos, and you can submit missing repos. You can also find some of them on my…

Optimizing the Metal pipeline to maintain 120 FPS in GPUI

Zed feels smoother than ever with today's release of 0.121, thanks to a series of optimizations that began on the kitchen table of popular streamer Theo Browne . In an excellent video following our open source launch, Theo gave a bunch of great feedback, but what really…

Zed is now open source

We hope you'll join us in our mission to fundamentally advance software collaboration.

A from-scratch tour of Bitcoin in Python

I find blockchain fascinating because it extends open source software development to open source + state. This seems to be a genuine/exciting innovation in computing paradigms; We don’t just get to share code, we get to share a running computer, and anyone anywhere can use it in…

Exploration Strategies in Deep Reinforcement Learning

[Updated on 2020-06-17: Add “exploration via disagreement” in the “Forward Dynamics” section . Exploitation versus exploration is a critical topic in Reinforcement Learning. We’d like the RL agent to find the best solution as fast as possible.…

The Multi-Armed Bandit Problem and Its Solutions

The algorithms are implemented for Bernoulli bandit in lilianweng/multi-armed-bandit . Exploitation vs Exploration The exploration vs exploitation dilemma exists in many aspects of our life. Say, your favorite restaurant is right around the corner. If you go there every day, you…