AI News Today

<pre style='white-space:pre-wrap;width:81ex'>refactor(agents): close embedded attempt terminal outcomes (#113201) * refactor(agents): close embedded attempt outcomes * test(agents): align trajectory terminal fixture * refactor(codex): share attempt failure type * fix(codex): retain projector result contract * test(agents): keep terminal projections test-only * chore(plugin-sdk): refresh attempt terminal baseline</pre>

Article URL: https://www.meyo.life/skill Comments URL: https://news.ycombinator.com/item?id=49030664 Points: 2 # Comments: 0

https://blog.seaticket.ai/how-to-resolve-software-deployment-issues-with-seaticket-8cff551ab44e Comments URL: https://news.ycombinator.com/item?id=49030661 Points: 1 # Comments: 0

We spoke with several cybersecurity researchers, who look for unknown vulnerabilities and develop tools to exploit them, about how OpenAI’s and Anthropic’s guardrails affect their work.

I wrote an interactive introduction to IEEE-754 floating point and would love some feedback. My main issue with existing treatments of this topic (and internet tutorials, broadly) is that they ask too little of the reader. I wanted to structure this around a sequence of questions for you to answer, using interactive widgets and optionally an LLM tutor/grader. Basically: you progress in the tutorial by demonstrating your understanding. The feedback I'm looking for is mostly around structure. Does

How do we get one of these? ScienceAlert stories are written, fact-checked, and edited by humans, never generated by AI. Don't miss a story, subscribe here.

Article URL: https://github.com/kdeps/kdeps Comments URL: https://news.ycombinator.com/item?id=49029939 Points: 2 # Comments: 0

Article URL: https://github.com/kdeps/turo Comments URL: https://news.ycombinator.com/item?id=49029919 Points: 3 # Comments: 2

As more details emerge about OpenAI's cyberattack against Hugging Face, lawmakers are taking an interest. PLUS: Meta's cynical new ad + "pervert glasses" damage control

<h2 id="part-1-from-one-support-bot-to-a-framework">Part 1: From one support bot to a framework</h2> <p>At Grab, AI agents have evolved from interesting team prototypes into production services used every day by millions of merchants, drivers, and consumers. Today, more than 500 services run on our internal agent framework, over 50 Model Context Protocol (MCP) servers are registered on our remote MCP framework, and a single Large Language Model (LLM) gateway fronts every model call across the co

Long-horizon execution in Large Language Models (LLMs) remains unstable even when high-level strategies are provided. Evaluating on controlled algorithmic puzzles, we demonstrate that while decomposition is essential for stability, extreme decomposition creates a “no-recovery bottleneck”. We show that this bottleneck becomes critical due to highly non-uniform error distribution, where consistent errors on a few “hard” steps become irreversible. To address this, we propose Lookahead-Enhanced Atom

Claude Opus 5 from Anthropic is now available on AI Gateway. Opus 5 improves on previous Opus models for long-horizon agentic coding, handling multi-file features, larger refactors, and end-to-end feature work, and completing full tasks rather than leaving stubs or placeholders. Opus 5 is effective at low and medium effort, which produce quality at a fraction of the tokens and latency of higher settings. Vision is stronger on charts, documents, diagrams, and UI replication, and Opus 5 makes effe

Attack models gonna attack

Article URL: https://github.com/ojki74/hallucinatoff Comments URL: https://news.ycombinator.com/item?id=49029558 Points: 2 # Comments: 0

Microsoft AI released two new in-house models into public preview on Wednesday — MAI-Image-2.5-Pro , its highest-fidelity image generator to date, and MAI-Voice-2-Flash , a speech model built for high-volume enterprise workloads — while publishing production data that amounts to the company's most aggressive argument yet that it can power its own products without leaning on OpenAI's frontier models. The announcement, made by Microsoft AI's Superintelligence team , lands roughly a year after the

<p> Your people, your agents, your project — all in one place </p> <p> <a href="https://www.producthunt.com/products/buzz-3?utm_campaign=producthunt-atom-posts-feed&utm_medium=rss-feed&utm_source=producthunt-atom-posts-feed">Discussion</a> | <a href="https://www.producthunt.com/r/p/1205055?app_id=339">Link</a> </p>

<p><strong><a href="https://martinalderson.com/posts/huggingface-openai-exploit/">The first known runaway AI agent - or a very bad marketing stunt?</a></strong></p> Martin Alderson's commentary on the <a href="https://simonwillison.net/2026/Jul/22/openai-cyberattack/">OpenAI accidental cyberattack against Hugging Face</a> includes a couple of details I hadn't considered.</p> <p>First, Hugging Face offers a truly rich target if you're trying to find potential vulnerabilities that require executin

Comments

Shashua has been invited to take the chairman of the board seat.

This is a small quality of life project i built for myself and using it for 3+ months now, decided to fork it out and publish as a skill separately. Instead of seeing terminals filled with code and tool calls, seeing a book quote felt like a breath of fresh air considering all of us are glued to it day and night. Hope the random quotes that pop up in your terminals lets you have a moment of pause or happiness every now and then while the agents keep churning out stuff for us. Please feel free to

Just wondering about this, because I don't think I've seen it said plainly. Hugging Face is the place where all us home-labbing AI enthusiasts get our models from. Aka, at the end of the day, OpenAI's likely mortal enemy. It feels like no one is mentioning this point, am I missing something? Comments URL: https://news.ycombinator.com/item?id=49028571 Points: 5 # Comments: 2

Article URL: https://github.com/highflame-ai/ai-factory Comments URL: https://news.ycombinator.com/item?id=49028457 Points: 1 # Comments: 0

The enemy of my enemy is my friend

Two weeks after debuting its more naturalistic GPT-Live audio AI model with full-duplex capabilities (listening and speaking at the same time), OpenAI is bringing it directly into developer workflows. The company announced that GPT-Live now powers the ChatGPT desktop application on macOS and Windows, integrating directly with agentic systems like Codex and ChatGPT Work (which are separate experiences available in the ChatGPT desktop app). When OpenAI initially launched GPT-Live on July 8, 2026,

Amazon is launching an update to its Alexa Plus assistant that will allow it to connect to smart home devices in new ways. With the update, which is currently in preview, Alexa Plus can link up with tech from Bosch, Delta, Ecovacs, iRobot, Yale Home, Whirlpool, Tapo, Eufy, and others, while automatically routing requests to […]

I built Skim because I was super dissatisfied with all existing Windows clients. I mean, it's not a rocket science, there should be some lightweight modern solution, isn't it? So I decided to create a client that just, you know, MAKES SENSE and has design younger than dinosaurs. Key points: * Obsessed with installer size: it should be under 5Mb. * BYOK AI. Cowriter, agentic search, summarization and other AI stuff is powered by your own key for Anthropic/Openrouter, or you can connect your own O

A new bill would let the US government order the shutdown of AI models that pose a public threat.

Microsoft has seemingly abandoned quality and craftsmanship a long time ago. But more recently they are rushing to force-feed us unfinished AI and web apps just so they can stay in the AI hype race. Call it teams, copilot, LinkedIn I've never liked anything that comes out of Ms lately, yet our company defaults to MS. At what point should we tell our employers to say no to MS treating paying customers as unpaid bulk beta testers? Comments URL: https://news.ycombinator.com/item?id=49027955 Points:

Article URL: https://prove-ai.github.io/agentpulse/ Comments URL: https://news.ycombinator.com/item?id=49027928 Points: 1 # Comments: 0

The incident is a first and signals a seismic shift in cybersecurity. The post OpenAI Agent Breaks Free and Hacks Hugging Face appeared first on SingularityHub .

ChatGPT Business and Enterprise users lack the standard chat export option, so third-party utilities like scrapemychats fill the gap

Article URL: https://github.com/Rentheria/llm-budget-cap Comments URL: https://news.ycombinator.com/item?id=49027830 Points: 1 # Comments: 0

Holiday's robot features a wheel-based mobile base, an upper body for dexterous manipulation, and hot-swappable batteries. The post Holiday Robotics raises $105M for its FRIDAY wheeled humanoid appeared first on The Robot Report .

OpenTrust is an open-source TypeScript SDK that collects privacy-preserving browser trust signals. It focuses on signals, not decisions: - browser automation detection - virtual camera detection - passive liveness signals The goal is to provide an additional trust layer for applications in a world where browsers may increasingly be controlled by automation and AI agents. Comments URL: https://news.ycombinator.com/item?id=49027707 Points: 1 # Comments: 0

<p> Cloud for your AI Agent's tools and functions </p> <p> <a href="https://www.producthunt.com/products/velane?utm_campaign=producthunt-atom-posts-feed&utm_medium=rss-feed&utm_source=producthunt-atom-posts-feed">Discussion</a> | <a href="https://www.producthunt.com/r/p/1204964?app_id=339">Link</a> </p>

AMD is challenging its chipmaker rival with a new rack-scale system that will start shipping to customers later this year.

Before agents entered the scene, the answer was obvious: A developer "writes" or "creates" software. While software craftsmen mourned the death of their craft, reality moved on. In February 2025, Andrej Karpathy ingeniously coined the term "vibe coding" (which hit Merriam-Webster as a slang term shortly after) to describe the shift from hand-writing code to steering the agent. But the honeymoon didn't last. Amateurs quickly ruined the vibe when their weekend projects got routinely annihilated up

In the AI-funding frenzy, many startups are raising back-to-back rounds at ever-increasing valuations — but even by those standards, Corgi stands out.