AI News Today

Comments

<p>Race to develop ‘embodied AI’ focuses on creating dextrous hands to transform humanoid robots from gimmicks into useful products</p><p>Human hands – nimble, nerve-filled appendages that are the most flexible part of the human skeleton – are exceptionally complex. Many tasks that most people can do largely without thinking, from tying a pair of shoelaces to buttoning up a shirt, in fact require a complex set of neurological instructions and precise choreography. In thousands of years of human

Article URL: https://www.clusy.io Comments URL: https://news.ycombinator.com/item?id=48799415 Points: 1 # Comments: 0

Article URL: https://noozra.com/articles/the-llm-big-bang Comments URL: https://news.ycombinator.com/item?id=48799386 Points: 2 # Comments: 0

Article URL: https://www.readtheline.ca/p/al-vigier-canadas-ai-strategy-shouldnt Comments URL: https://news.ycombinator.com/item?id=48799256 Points: 3 # Comments: 1

Article URL: https://github.com/PJHkorea/egregore-core-jax Comments URL: https://news.ycombinator.com/item?id=48799235 Points: 2 # Comments: 1

Speech-only spoken language models (SLMs) lag behind text and text-speech models in performance, with recent discrete autoregressive (AR) SLMs indicating significant computational and data demands to match text models. Since discretizing continuous speech for AR creates bottlenecks, we explore whether continuous diffusion (CD) SLM is more viable. To quantify the SLMs linguistic quality, we introduce the phoneme Jensen-Shannon divergence (pJSD) metric. Our analysis reveals CD SLMs, mirroring AR b

Language models play a central role in automatic speech recognition (ASR), yet most methods rely on text-only models unaware of ASR error patterns. Recently, large language models (LLMs) have been applied to ASR correction, but introduce latency and hallucination concerns. We revisit ASR error correction with compact seq2seq models, trained on ASR errors from real and synthetic audio. To scale training, we construct synthetic corpora via cascaded TTS and ASR, finding that matching the diversity

Safety policies define what constitutes safe and unsafe AI outputs, guiding data annotation and model development. However, annotation disagreement is pervasive and can stem from multiple sources such as operational failures (annotators misunderstand or misexecute the task), policy ambiguity (policy wording leaves room for interpretation), or value pluralism (different annotators hold different perspectives on safety). Distinguishing these sources matters. For example, operational failures call

Look past the data-center megadeals and the bubble headlines, because something quieter happened this week: AI actually delivered. A Nobel laureate joined the AI-for-science push, the open-model frontier got cheaper again, and the first hard evidence landed that an AI tutor can beat the classroom. These are the wins that did not make the noise.

Article URL: https://mattmccormick.ca/we-re-all-managers-now-my-journey-into-ai-assisted-development/ Comments URL: https://news.ycombinator.com/item?id=48799212 Points: 3 # Comments: 0

I built an open source terminal app that transparently wraps your shell so you can have access to AI whenever you need it, while it stays completely out of your way the rest of the time. Bring your own AI agent (Claude and Codex are supported out of the box) with auto-configured MCP access to the contents of your terminal and scrollback and the ability to suggest input that you can approve passing to your shell as if you had typed it yourself. The hardest part technically was getting it to be tr

A respected engineer has informed me that the terms of service of most if not all AI outfits mandate that anything you submit becomes their property, and he said that already he knows of 2 instances of startup ideas being stolen away by the AI outfits. He did not name names. Comments URL: https://news.ycombinator.com/item?id=48798971 Points: 3 # Comments: 2

<pre style='white-space:pre-wrap;width:81ex'>fix(openai): bound realtime voice websocket payload at 16 MiB (#99450) * fix(openai): bound realtime voice websocket payload at 16 MiB * test(openai): cover realtime voice payload cap --------- Co-authored-by: Peter Steinberger <steipete@gmail.com></pre>

<pre style='white-space:pre-wrap;width:81ex'>fix(agents): trim media tools in lean mode (#88881) * fix(agents): trim media tools in lean mode * fix(agents): preserve lean tool allowlists * fix(agents): preserve lean tool overrides * fix(agents): enforce lean policy at harness boundary * fix(agents): preserve lean override provenance * test(agents): satisfy lean harness contracts * docs(changelog): record lean local-model tool trimming --------- Co-authored-by: Peter Steinberger <steipete@gmail.c

Comments

<p> Trading desk for LLM calls </p> <p> <a href="https://www.producthunt.com/products/auriko?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/1188837?app_id=339">Link</a> </p>

If you check the newest submissions right now (https://news.ycombinator.com/newest), ten out of 30 are already dead, some only sent five minutes ago. On the second page, it's 9. On the third, 11. It seems unlikely that many people are downvoting them so quickly, and for most I can't see why they should be dead. I know that there's websites known to publish only AI slop which are probably blacklisted here, but it seems unlikely that a third of the submissions are about them. There's some automati

Article URL: https://ai-2027.com/?agi=true Comments URL: https://news.ycombinator.com/item?id=48798738 Points: 2 # Comments: 2

<p>Minecraft terrain, roguelike dungeons, and No Man's Sky planets all come from the same place: rules, not hand built content. If you strip procedural generation down, nearly everything ships on three algorithm families.</p> <p><strong>Noise functions.</strong> Perlin and simplex noise give you smooth randomness, which is exactly what terrain wants. Sample 2D noise for a heightmap, layer a moisture map on top, threshold it, and you have biomes. Almost every survival game terrain starts here.</p

<p>When people hear "an app that writes books with AI," they picture one giant prompt and a wall of purple prose. That's the demo version. The moment you try to ship a real reading-and-writing product, you find out the generation step is the easy part. The hard part is everything wrapped around it — and it's a genuinely interesting systems problem.</p> <h2> A book is a state machine, not a prompt </h2> <p>The first thing that breaks the naive approach is length. Ask a model for "a novel" and you

<p>Today started with a simple goal: finish a feature on an AI-based project I'm building.</p> <p>That didn't happen.</p> <p>Instead, I spent almost five hours debugging.</p> <p>At first, I was convinced the database was causing the problem. After fixing that, another issue appeared. Then I thought it was the queue. Then the AI API. Then the fetch logic.</p> <p>The interesting part wasn't the bugs—it was how I was approaching them.</p> <p>I stopped trying random fixes and started tracing the req

<p>Quick disclosure before anything else: I'm on the team behind this. Not a neutral reviewer, just someone who got tired of the same thing you probably did.<br> So a few of us built AntSeed. It's an open market for AI inference. Anyone can sell intelligence, anyone can buy it. Your request goes straight to whichever provider fits, cheapest, fastest, or the one with the best track record, instead of whoever your subscription picked for you months ago.<br> The part that actually convinced me wasn

Most Americans don't trust AI. It's proven that it doesn't know what safe toppings for pizza are. People don't even want to listen to AI music. But none of that matters for some of America's wealthy, who are turning to AI to teach their kids instead of traditional schools. Companies like Forge Prep and Alpha […]

<p>Eighteen months ago, agent-driven ad buying was a keynote topic. Today it is two competing standards stacks with shipped software, published by two different organizations that are openly not coordinating.</p> <p>If you build adtech, both are about to show up in your pipes. Here is the short version.</p> <h2> Stack one: IAB Tech Lab's AAMP </h2> <p>AAMP (Agentic Advertising Management Protocols) launched in January 2026 and hit 2.0 in June. The philosophy: build on the rails that already move

<pre style='white-space:pre-wrap;width:81ex'>feat(chat-ui): redesign composer bottom bar with split Talk control and reasoning-effort chip (#100445) * feat(chat-ui): redesign composer bottom bar with split Talk control and reasoning-effort chip Removes the toolbar divider so the composer frame is the only chrome, merges the Talk toggle and Talk settings gear into a mic+caret split control, shows the effective reasoning level as a subtle pill next to the model name (accent-washed only on session

<pre style='white-space:pre-wrap;width:81ex'>fix(release): package legacy candidates without AI workspace (#100469)</pre>

<pre style='white-space:pre-wrap;width:81ex'>fix(slack): prefer native status by default (#100462) Co-authored-by: Peter Steinberger <steipete@openai.com></pre>

I am very paranoid about sharing potential AI research with e.g. Claude [Code] or ChatGPT/Codex. I believe that any company is essentially a paperclip optimizer that will do whatever it takes to win over competition. AI companies have access to the IP of millions of AI researchers and AI startups who are in direct competition with them. If they can use this data to squash competition (either competition from the same researchers or from others), I believe that they will use if eventually (if not

<p> AI Music Tutor that gives real-time feedback on your playing </p> <p> <a href="https://www.producthunt.com/products/soundgate-guitar-app?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/1188822?app_id=339">Link</a> </p>

If you have a book idea, Youbooks AI Non-Fiction Book Generator can make it a reality with this lifetime subscription.

Article URL: https://www.windowslatest.com/2026/07/05/microsoft-365-just-got-a-price-hike-over-continuous-innovation-but-copilot-is-the-ai-tax-on-businesses/ Comments URL: https://news.ycombinator.com/item?id=48798330 Points: 19 # Comments: 8

Article URL: https://github.com/irenee28/akm-clr-governance-layer Comments URL: https://news.ycombinator.com/item?id=48798328 Points: 2 # Comments: 0

<!-- SC_OFF --><div class="md"><p>Hi everyone, I'm currently working on a framework to evaluate the security of LLM applications and AI agents, and I've been stuck on one part for a while.</p> <p>Most red-teaming frameworks rely on an LLM to generate adversarial prompts. My question is more about <strong>which model to use</strong>.</p> <ul> <li>Which <strong>closed-source</strong> models would you recommend for generating high-quality attacks?</li> <li>Which <strong>open-source</strong> models

Article URL: https://github.com/bharadwaj-pendyala/sidenote Comments URL: https://news.ycombinator.com/item?id=48797739 Points: 1 # Comments: 0

https://xcancel.com/plannotator/status/2073845811847512372?s... https://github.com/backnotprop/plannotator Plannotator Code Review works for any local or PR changes, jj, and p4. You can conduct a code review on your own or integrate it with your agents and submit feedbacks to create an automated iteration loop. Guided reviews works just like linear, except it also summarizes all file changes for you. The most important code changes are surfaced to the top, provided with a semantic overview & gro

Article URL: https://github.com/SakanaAI/fugu Comments URL: https://news.ycombinator.com/item?id=48797562 Points: 4 # Comments: 0

Article URL: https://www.euronews.com/travel/2026/07/03/tripadvisor-ai-summaries-give-glowing-reviews-to-dangerous-hotels-consumer-watchdog-finds Comments URL: https://news.ycombinator.com/item?id=48797529 Points: 25 # Comments: 9

Hi HN, This is ProofTree, and in TLDR: it is a platform where you can chat with an AI to do math, the way you already do, with context-awareness and knows how you personally do math, connected to the fora of other live human mathematicians. So, when you are talking to an AI about a proof at 3 am, and you need a human expert looking at it, or a discussion around it, like you do on StackExchange, this is it! You need an approved email account to use it, so please write here if you would live to tr

Aieeee! ScienceAlert stories are written, fact-checked, and edited by humans, never generated by AI. Don't miss a story, subscribe here.