AI News Today

Agentic era has left me with already half a dozen vibe coded helper apps, most run with ´bun run dev´ or the npm equivalents, occasionally Python. Instead of zooming around terminal, I decided to vibe code one more thing: A local dashboard to start and stop all the projects and view the ouput. Finessing the configuration yaml is delegated to a SKILL.md so I can just ask Claude or Codex to set the stuff up. Works and saves time! It's Bun + Svelte, MIT licensed, binds to localhost with no auth (it

Enterprise companies are running AI agents ahead of the controls needed to manage them — and they deployed that way knowingly. That is the central finding from VentureBeat Research's June survey of 573 technical leaders at companies with 100 or more employees, fielded across five parallel surveys of the agentic stack. Enterprises are now retrofitting to catch up with their own standards, and they are budgeting for it: Roughly six in 10 enterprises plan to switch or add vendors in each of five co

We build an autonomous data science agent around DeepAnalyze-8B and run it end to end. We prepare a stable Colab runtime, install the machine-learning dependencies, and load the tokenizer and model in 4-bit mode to fit limited GPU memory. We add a sandboxed execution environment that lets the model generate Python, run it safely, observe results, and continue in an agentic loop. We then hand the agent a multi-file e-commerce workspace and let it clean, join, analyze, visualize, and summarize the

Advocates are demanding that the Department of Homeland Security release bodycam footage of the fatal shooting of Lorenzo Salgado Araujo, a Mexican immigrant who was killed by ICE officers in Houston during a traffic stop earlier this week. But DHS claims the agents involved in the shooting weren't wearing body cameras because of the lengthy […]

<p> Shared memory across all apps and LLMs. In Claude </p> <p> <a href="https://www.producthunt.com/products/unabyss?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/1193175?app_id=339">Link</a> </p>

Article URL: https://www.mccormick.northwestern.edu/news/articles/2026/07/ai-gets-a-cerebellum/ Comments URL: https://news.ycombinator.com/item?id=48863998 Points: 3 # Comments: 0

<!-- SC_OFF --><div class="md"><p>Zhang et al. paper's introducing the SA-MDP framework (2020) (state adversarial MDP) argues that an attack using the critic network (V(s)) is expected and supposed to produce a weaker attack than an attack using the actor network (pi(s)) itself to generate perturbation on agent observations. A claim supported by their empirical results using different single-agent simulation environment.</p> <p>However, I'm consistently finding the opposite when comparing both a

Article URL: https://trace-api-ixv6o.ondigitalocean.app/api/portal/webhook Comments URL: https://news.ycombinator.com/item?id=48863824 Points: 2 # Comments: 0

Open source AI is booming, according to Hugging Face CEO Clem Delangue. The company has grown into something like a GitHub for AI in recent years, where AI builders can share and download open models and datasets, now used by roughly half the Fortune 500. Delangue has seen the same story play out again and again: companies start […]

Using a local LLM for coding agents is complicated, so I built a Cursor + VS Code plugin that makes it easy. It automatically profiles your machine, picks a model that works for your machine, then runs a local server, sets up a Cloudflare Quick Tunnel and walks you through connecting it to Cursor. In the end, you work like you normally do in Cursor and just choose "Local Motion" as your model in the model picker dropdown. Comments URL: https://news.ycombinator.com/item?id=48863722 Points: 1 # Co

Comments

According to OpenAI, GPT-5.6 Sol independently fine-tuned the smaller Luna model, triggered by a single "fairly under-specified prompt." In OpenAI's internal RSI benchmark for recursive self-improvement, Sol scores 16.2 points higher than GPT-5.5. OpenAI believes the "automated researcher" is within reach. The article OpenAI's GPT-5.6 Sol autonomously post-trained the smaller Luna model with a "fairly underspecified prompt" appeared first on The Decoder .

Comments

Enterprise AI teams are giving agents more freedom at the same moment their confidence in automated testing is collapsing. Half of enterprises have deployed an AI agent or LLM feature that passed internal evaluations and yet still caused a customer-facing failure — one in four more than once — according to the June 2026 VB Pulse survey of 157 qualified enterprise respondents at companies with 100 or more employees. The sample is self-selected rather than a probability sample, so the findings sho

Large language model (LLM) training workloads increasingly run into GPU memory limits before compute is fully used. Model weights, gradients, optimizer states,...

Article URL: https://github.com/PJHkorea/value-system-kernel Comments URL: https://news.ycombinator.com/item?id=48863292 Points: 3 # Comments: 1

ive been seeing a recurring claim that open (weight) models 6 months behind the frontier are good enough for the majority of ‘work’. if you've had a concrete task in the last month where GLM/DeepSeek/Kimi/Qwen failed and Opus/Fable/GPT succeeded (or vice versa!), please share Ill provide a template as ive also frequently seen posters complain about a lack of context: - task (short and specific) - cheap model tried & how it failed - frontier model & did it actually succeed - would frontier-1 have

Researchers found a simple solution for extending the lifespans of LEDs made from glowing microscopic particles called quantum dots.

Hey HN! We were part of YC S24, ended up pivoting around. Made dictation that worked really well, and decided to go all in on agentic. We've done two major things. 1. Scribe is a revamp of our product. Continually learns your style (can get into details if comments want). TLDR; we fingerprint based off your edits, and make dictation sound more and more like you. Post-trained Llama 3.1 8b. 2. Dictation is completely free. Unlimited unpersonalized dictation for everyone. Comments URL: https://news

GPT-5.6 Sol ships with five reasoning levels from "Light" to "xhigh," plus "Max" and "Ultra" modes that deploy multiple sub-agents in parallel. OpenAI's Vaibhav Srivastav recommends starting low and only scaling up when needed. The article OpenAI staffer maps out which of GPT-5.6 Sol's five reasoning levels fits which task complexity appeared first on The Decoder .

<p> Approve AI requests from your lock screen </p> <p> <a href="https://www.producthunt.com/products/pushary?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/1193118?app_id=339">Link</a> </p>

Article URL: https://hal9.com/articles/the-2025-2026-evolution-of-generative-spatial-ai Comments URL: https://news.ycombinator.com/item?id=48863008 Points: 1 # Comments: 0

Atlas will be shutting down as OpenAI introduces ChatGPT Work.

Article URL: https://www.bloomberg.com/news/articles/2026-07-10/big-tech-doubles-debt-load-to-350-billion-in-ai-spending-spree Comments URL: https://news.ycombinator.com/item?id=48862847 Points: 2 # Comments: 1

As the AI boom boosts demand for RAM, SK Hynix - one of the world's biggest suppliers of memory chips - launched on Wall Street Friday. The South Korean chipmaker opened at $170 per share and raised $26.5 billion, surpassing Alibaba's record as the largest debut of a foreign company, according to reports from Associated […]

The AI chip boom just produced its biggest Wall Street moment yet. Now SK Hynix and Samsung are being asked to build U.S. factories.

HydroShear, a new physics-based simulator, teaches robots how to use their sense of touch to perform complex manipulation tasks, in a way that transfers seamlessly to the real world.

GenUI lets AI agents build real, interactive SwiftUI interfaces for iOS and macOS instead of just replying with text. PS: still needs some work for design quality Comments URL: https://news.ycombinator.com/item?id=48862581 Points: 1 # Comments: 0

<blockquote cite="https://youtu.be/v4vkwUf4AMw?t=2427"><p>The reality is to make augmented reality glasses, you need to put a camera next to your eyes that is continuously recording everything you see and processing that to put information over it.</p> <p>There is not another way around it. And there's certainly not a chip that can fit in the stem of a glasses that is both powerful enough and power miserly enough to do that in real time.</p> <p>You have to send that data to a cloud. You gotta do

Spotify is giving listeners control to fine-tune what gets surfaced for them in Release Radar - one of its most popular weekly playlists. The new options allow you to narrow the playlist to a specific genre, focus on artists that are new to you, and more. Listeners can choose from up to five options like […]

Building a production-ready RSS pipeline with Python, Docker, PostgreSQL, and Kestra The post I Built My Second ETL Pipeline. This Time, I Started Thinking Like a Data Engineer appeared first on Towards Data Science .

Handoff willingness, flexibility, proactivity, emotional responsiveness, and transparency help you build trustworthy AI chatbots that guide users well.

<pre style='white-space:pre-wrap;width:81ex'>fix(anthropic): delegate adaptive Claude CLI effort (#103815) * fix(anthropic): delegate adaptive CLI effort Strip static --effort args for adaptive runs so Claude Code resolves effort from its environment, settings, and model default. Preserve configured effort for off or absent thinking and replace it only for concrete OpenClaw levels. Fixes #103245 Co-authored-by: Dan Korotin <korotin.daniil@gmail.com> * fix(anthropic): make effort dispatch exhaust

<pre style='white-space:pre-wrap;width:81ex'>fix(embedded-agent-runner): treat the run's own no-op in-place session rewrite as benign (#91797) * fix(agents): accept byte-identical session rewrites Co-authored-by: gucasbrg <buruguo2000@163.com> * chore: keep changelog release-owned --------- Co-authored-by: Peter Steinberger <steipete@gmail.com></pre>

<p>We’ve all been there: you saw a link, a funny quote, or an important address on your screen a few days ago, but you can’t remember which app it was in. Was it WhatsApp? A random website? A fleeting notification?</p> <p>Big tech’s answer to this is usually cloud-based memory or "recall" features that stream your screen context to a server. That solves the memory problem, but it creates a massive privacy nightmare.</p> <p>When I previously built Pribado—a general-purpose, non-crypto key managem

Tencent is in talks to buy a majority stake in AI agent startup Manus at the same $2 billion valuation, according to the Financial Times, after Beijing blocked Meta's acquisition. Tencent sees overlap with its own agent plans, including for WeChat. U.S. firm Benchmark is not expected to take part. The article Tencent moves to buy majority stake in Manus after Beijing forced Meta to unwind its $2 billion deal appeared first on The Decoder .

<pre style='white-space:pre-wrap;width:81ex'>[AI-assisted] Tidy Android Sessions row affordances (#103108) * fix(android): clean up Sessions row visuals * chore(android): refresh native i18n inventory * fix(android): consolidate session sorting controls --------- Co-authored-by: IWhatsskill <284122573+IWhatsskill@users.noreply.github.com></pre>

<blockquote> <p><em>To be on the hook is to be the one who answers when something breaks. Every team carries a picture of where that hook hangs. This is about whether the picture was ever true.</em></p> </blockquote> <p>It’s the first question any skeptical engineering leader asks about this model. The agent implements from a spec the team agreed on. Something breaks in production. Who’s on the hook?</p> <p>The question assumes an answer that was never true to begin with.</p> <p>Before any of th

There are many ways to optimize code for GPUs. In this post, you’ll learn how kernel fusion can improve memory bandwidth and reduce kernel launch overhead,...

<pre style='white-space:pre-wrap;width:81ex'>feat(agents): derive a provider-declared default utility model when unset (#103769) * feat(agents): derive a provider-declared default utility model when unset When agents.defaults.utilityModel is not set, utility tasks (titles, progress narration) now use the primary provider's declared small model (modelCatalog.providers.<id>.defaultUtilityModel: OpenAI -> gpt-5.6-luna, Anthropic -> claude-haiku-4-5). Auth is inherent because derivation follows the