AI News Today

<!-- SC_OFF --><div class="md"><p>I just now posted a small research / ML interp study on symmetries inside the neural nets for an open source Go-playing program that I maintain ("KataGo"). The rules of Go are completely symmetric under rotation/reflection, but such symmetry is not enforced in the models - the only thing we do for that is stochastic 8-fold data augmentation during training, randomizing the spatial orientation of each batch. </p> <p>To what degree do superhuman-strength Go-playin

OpenAI's refutation of the Unit Distance Conjecture has sparked a wave of AI-assisted advances in mathematics. Fields Medal winner Timothy Gowers says GPT 5.6 Pro solved two problems he had spent considerable time working on, each on its first attempt. He warns of the "possible destruction of mathematical culture" if mathematicians stop building the expertise needed to understand such results. Others simply see AI as a productivity tool. The article AI keeps cracking unsolved math problems, and

Uber has partnered with — and in some cases made direct investments in — about 30 autonomous vehicle companies over the past two years. Here's the list and the latest on the partnerships.

How far does it go? ScienceAlert stories are written, fact-checked, and edited by humans, never generated by AI. Don't miss a story, subscribe here.

Most coding agents treat prompt construction like retrieval: gather more files, add more context, hope the model figures it out. But that approach breaks down fast. As context grows, irrelevant code competes for attention, and when the window fills, agents start compressing their own memory—often mid-task. What looks like “forgetting” is usually just degraded context. This article explores a different approach: treating prompt construction like a compiler that decides what to keep, what to reduc

Google rolls back Google Earth Nano Banana 2 addition one day after launch.

Article URL: https://blog.sparsh.dev/how-to-prompt-llms/ Comments URL: https://news.ycombinator.com/item?id=49134770 Points: 3 # Comments: 2

A field report from OpenAI and academic partners shows coding agents can modernize neglected research software, with speedups of up to 60x. But the systems are "eloquent, convincing, and confidently wrong in ways that are easy to miss," participants say. The effort shifts from writing code to the time-consuming work of verifying scientific correctness. The article AI coding agents can modernize research software but can't judge if the science is right appeared first on The Decoder .

<p> The fastest workflow for developing with AI </p> <p> <a href="https://www.producthunt.com/products/agent-manager?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/1212310?app_id=339">Link</a> </p>

<p> A Claude Code alternative for people who avoid the terminal </p> <p> <a href="https://www.producthunt.com/products/lumichats-offline?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/1212303?app_id=339">Link</a> </p>

A security researcher has demonstrated a worm-like attack on Microsoft Copilot for Word: invisible prompt injections hidden in documents spread automatically into new files every time they're reused. Microsoft confirmed the issue but failed to fix it after 144 days and two attempts. The article A security researcher built a self-spreading worm that hides inside Word docs and hijacks Microsoft Copilot appeared first on The Decoder .

Hi HN. Wyro is a visual backend builder that compiles to a plain TypeScript repo you own outright. The motivation: AI app builders got good at generating apps and stayed bad at letting you leave. Frontend code you can always lift out. The backend is the hostage — their runtime, their DB, their deploy target. How it works: you wire blocks on a canvas (HTTP routes, tables, auth, queues, cron, multi-step pipelines) into a graph. That graph is an IR, and the compiler lowers it to a conventional Node

<p>Google’s April 2026 Gemini Drop broadens how the Gemini app connects across desktop and mobile experiences. The update adds a native macOS app, integrates NotebookLM into Gemini conversations and introduces new ways to organize chats. It also sits within Google’s wider push toward agent-like task automation, including Gemini Agent for eligible Ultra subscribers.</p> <p>The company’s <a href="https://blog.google/innovation-and-ai/products/gemini-app/gemini-drop-april-2026/" rel="noopener noref

<p>Hello Devs 👋</p> <p>If you've looked at AI code review tools recently, you've probably seen the term <strong>"context-aware code review"</strong> everywhere.</p> <p>Every tool seems to use it.</p> <p>Some talk about repository awareness. Others mention codebase understanding. A few claim they understand architecture and engineering best practices.</p> <p>After spending some time reading docs, testing different tools, and comparing how they review pull requests, I realized something:</p> <p>M

Article URL: https://frontierroles.com/ Comments URL: https://news.ycombinator.com/item?id=49134249 Points: 1 # Comments: 0

<pre style='white-space:pre-wrap;width:81ex'>fix(agents): settle user turn persistence before reconciliation (#117393)</pre>

<p>A renewal agent can call the CRM, email, calendar, support, document search, and contract systems. On its first run, it asks every system for everything related to one customer.</p> <p>The result looks thorough: hundreds of CRM fields, years of ticket history, complete email threads, and the same contract attached to several messages. But the agent now has to spend time and tokens deciding which records matter before it can make progress on the renewal.</p> <p>The agent is well connected. It

<p>AI coding agents are strangely lopsided. Hand one a well-scoped task and it writes better code, faster, than I do. Hand it a <em>large</em> feature, the kind that takes days and touches a dozen files across several layers, and the failure mode changes. The context window fills up and quality quietly degrades (AKA <strong>context rot</strong>). A fresh session, missing the history, confidently invents an API that never existed. And even when the code is good, you're left staring at a 3,000-lin

<pre style='white-space:pre-wrap;width:81ex'>fix(codex): guide recovery from unimported agent auth (#116807) * fix(codex): diagnose unimported agent auth Add agent-scoped Codex migration guidance and expose a validated --agent target so non-default credentials cannot be imported into the default agent store. Fixes #116763 * fix(codex): surface auth migration before route planning * fix(codex): scope auth migration recovery * fix(codex): preserve API key auth fallback * test: repair memory import

A Beginner's Guide to Working with Claude Design • 5 Best AI Tools for Data Analysis You Should Try in 2026 • 5 Books That Will Deepen Your Understanding of Large Language Models • Is KimiClaw a Useful Tool?

A vital survival skill in a warming world. ScienceAlert stories are written, fact-checked, and edited by humans, never generated by AI. Don't miss a story, subscribe here.

A hybrid LLM application pattern that combines a predefined workflow with adaptive agent behavior The post Put the Agent Inside the Workflow appeared first on Towards Data Science .

I built a skill for building agents that helped me ship agentic products within days that otherwise took several weeks and broke often in production. It is inspired by gstack. Comments URL: https://news.ycombinator.com/item?id=49134004 Points: 1 # Comments: 0

<pre style='white-space:pre-wrap;width:81ex'>fix(agents): avoid provider discovery when cache pruning is off (#117396)</pre>

Article URL: https://github.com/MiskaKan/thinair Comments URL: https://news.ycombinator.com/item?id=49133864 Points: 2 # Comments: 0

Reddit may still be considering ending its licensing deal with Google.

Article URL: https://github.com/pochenai/nano-llm-posttraining Comments URL: https://news.ycombinator.com/item?id=49133851 Points: 2 # Comments: 0

<!-- SC_OFF --><div class="md"><p>I am trying to learn concepts like On Policy Distillation (OPD), On Policy Self Distillation (OPSD) and how do they compare to RL algorithms like GRPO.</p> <p>There are a lot of papers on this, but because of limited compute I cannot try these papers out and learn them by implementing them myself.</p> <p>If someone here has worked with these algorithms and their implementation on SLMs (something that can fit a consumer grade GPU like Nvidia RTX 4090 or 5090), ca

Article URL: https://freeaipromptgen.com/ Comments URL: https://news.ycombinator.com/item?id=49133672 Points: 10 # Comments: 4

<p>New AI models, advanced robots and specialty computer chips are unsettling markets and the US tech industry</p><p>Over the past month, a series of advancements in China’s artificial intelligence, chip manufacturing and robotics technologies have rattled financial markets, caused divisions among US tech moguls and left the Trump administration scrambling to respond.</p><p>Silicon Valley has long pointed to China’s tech industry as a competitive threat and used its growth as rationale for why U

Agent Platform's evaluation service is now generally available, providing developers with a unified engine to measure agent quality consistently across local development experiments and live production traffic. You can evaluate agents using over 20 pre-built metrics, DeepMind-backed adaptive rubrics, or custom code-based and LLM-as-a-judge metrics stored in a centralized, versioned registry. The service integrates directly into existing workflows via the Agent Platform SDK, agents-cli, and ADK,

To prevent context window bloat and reduce token consumption, Genkit Go introduces Agent Skills based on a progressive disclosure architecture. Developers can package specialized instructions, scripts, and references into modular SKILL.md bundles where only the frontmatter metadata is initially exposed to the agent's system prompt. When a task matches the skill's description, Genkit's middleware dynamically loads the full instruction body and associated assets, ensuring the model accesses precis

In July, large funding rounds, new approaches to robot training, and new robot models drew our readers attention. The post Top 10 robotics stories of July 2026 appeared first on The Robot Report .

<pre style='white-space:pre-wrap;width:81ex'>fix(tui): cancel buffered submissions on shutdown [AI-assisted] (#117331) * fix(tui): cancel buffered submissions on shutdown * refactor(tui): preserve callable submit coalescer * test(tui): run local PTY smoke against built CLI</pre>

Our weekly science news roundup. ScienceAlert stories are written, fact-checked, and edited by humans, never generated by AI. Don't miss a story, subscribe here.

Autonomous vehicle giant disturbs residents' sleep.

<p>A New York school district has paused plans to bring an android into the classroom. Is this our educational future?</p><p>What’s more harrowing than sending your child to school? Maybe <a href="https://en.wikipedia.org/wiki/Speed_(1994_film)">getting stuck on a city bus</a> that can’t drop below 55 miles per hour or it will explode, but that’s about it. Once your kid hits school age, you have to decide (if you’re lucky) between public and private education, vet the teachers, and neurotically

Plus: The FBI eyes AI-powered tech to detect future crimes, Russia charges Telegram’s founder, xAI sues to stop a state’s “nudification” ban, and the Democrats learn a lesson about getting scammed.

Article URL: https://polymyth.quest Comments URL: https://news.ycombinator.com/item?id=49133025 Points: 1 # Comments: 0

A removable chip lets hackers inspect their badge—and keep using it after Defcon.