AI News Today

Like a lot of folks, I'm trying to understand the AI reality vs. the hype cycle. I have my own experiences both using it and seeing it rolled out with a “do more AI” mandate at a huge tech company. Even looking here at discussions we have the full spectrum from engineers who say they don’t code anymore, they just manage agents, to engineers who say they find it basically useless because of how stupid it is. I’ve been looking for clues (I wanted to use the word “signals" there but I can’t because

Power is a defining constraint for AI factories. As AI workloads demand a full compute platform to serve them, each component of that platform must maximize...

For operators of large-scale AI factories, maximizing continuous output is essential for productivity. In massive-scale AI training, every GPU in the cluster...

Article URL: https://www.normaltech.ai/p/the-ai-as-normal-technology-view Comments URL: https://news.ycombinator.com/item?id=49715332 Points: 1 # Comments: 0

Article URL: https://newsletter.posthog.com/p/if-ai-writes-all-the-code-whats-left Comments URL: https://news.ycombinator.com/item?id=49715305 Points: 1 # Comments: 0

Article URL: https://www.youtube.com/watch?v=9tr7Mby62bo Comments URL: https://news.ycombinator.com/item?id=49715298 Points: 1 # Comments: 0

Comments URL: https://news.ycombinator.com/item?id=49715289 Points: 2 # Comments: 1

Article URL: https://intelligence.mts.now/compute Comments URL: https://news.ycombinator.com/item?id=49715276 Points: 2 # Comments: 0

Article URL: https://scoopkit.dev/ Comments URL: https://news.ycombinator.com/item?id=49715271 Points: 1 # Comments: 0

G5 Labs, an MIT Computer Science and Artificial Intelligence Laboratory (CSAIL) spinout, has emerged from stealth with $14 million in seed funding and an ambitious thesis: the next major programming language may not look like a programming language at all. The Boston-based startup is developing an abstraction layer for AI-native software development in which natural-language descriptions of what software should do are organized into a formal graph of intent, or ontology, that effectively…

Article URL: https://www.axios.com/2026/09/15/ai-trust-safety-anthropic-openai Comments URL: https://news.ycombinator.com/item?id=49715182 Points: 3 # Comments: 0

Lectra on September 15, 2026 announced the launch of Apogy, a cloud-based solution that embeds agentic artificial intelligence in fashion product development, the phase between garment design and production. Irish sportswear brand O'Neills and international fashion group Oniverse are among its early customers. Lectra describes Apogy as a single digital environment that brings together the data, processes and stakeholders involved in developing a product. According to the company, the software…

Anthropic expanded Claude for Small Business on September 15, 2026, bringing the small-business plugin to 43 workflows and adding 27 new integrations with tools including Shopify, Salesforce, TikTok, Atlassian, Zoom, Xero, Gusto, Square, Stripe, and Zapier. The release arrives with a fall schedule of free in-person workshops, partner webinars, and community-run training. New Workflows and How They Run Claude for Small Business runs in Claude Cowork, the desktop app where Claude works directly…

Sorry if this is a stupid question but... Problem: I boot opencode into a codebase. I ask it to do one thing, it has to call 40 tools to "remember" what the codebase looks like relevant to that. It has no memory so every fresh boot is re-explore. My understanding, as an ML n00b, is the GPU have KV states of massive matrices that are multi GB in RAM (e.g., 10-30GB) so dumping to disk and pushing over network ( his shoulders shuddered at the thought of egress fees ), is not great. So instead we ha

Autodesk previewed new agentic AI capabilities across its three industry clouds at Autodesk University 2026 in Las Vegas on September 15, 2026, saying the capabilities shape the next generation of Autodesk Assistant, with the expanded experience planned to begin rolling out in 2027. The preview spans Autodesk Forma for architecture, engineering, and construction; Autodesk Fusion for design and manufacturing; and Autodesk Flow for media and entertainment. Together, the new capabilities extend…

<!-- SC_OFF --><div class="md"><p>Prior Labs released their latest tabular foundation model, TabPFN-3.5 today.</p> <p>The model is top of both TabArena and BeyondArena and SOTA for 1M rows and up to 20k features</p> <p>It comes with:</p> <p>- TabPFN-3.5-Fast (in alpha): This one goes 6x faster than the base model<br/> - TabPFN-3.5-Thinking: you basically exchange compute for better accuracy with this one and it's via the API<br/> - TabPFN-3.5-Plus</p> <p>On BeyondArena, TabPFN-3.5 leads on text-

Prompt caching in Amazon Bedrock can cut input token costs by up to 90% when you repeatedly send the same context to foundation models. This post walks through six practical prompt caching scenarios using the Converse API: message content, system prompt, tool definition, mixed TTL, tenant isolation, and LangChain integration.

Manually tagging thousands of catalog products is slow and inconsistent. This walkthrough shows how to customize Qwen3-8B with supervised fine-tuning (SFT) and reinforcement learning with verifiable rewards (RLVR) on Amazon SageMaker serverless model customization, then deploy it for asynchronous inference to build a cost-efficient product tagging system.

Amazon SageMaker AI now offers instance preference lists for training and processing jobs. Specify an ordered list of up to five instance types, and SageMaker AI automatically launches on the first type with available capacity, eliminating manual retry loops and capacity-watching scripts.

As AI labs draw professors and students away from universities, society risks losing the expertise to govern what those labs build.

<img src="https://storage.googleapis.com/gweb-uniblog-publish-prod/original_images/Health_Header.gif">Explore this collection to see how experts and local leaders are using AI breakthroughs to ensure everyone can share the opportunity of AI.

<img src="https://storage.googleapis.com/gweb-uniblog-publish-prod/images/SocialShare_hD2gcAw.max-600x600.format-webp.webp">The true measure of AI is who it helps. Here’s how it’s impacting lives today. We're focused on key areas where advanced technology can help make extraordinary progress …

<img src="https://storage.googleapis.com/gweb-uniblog-publish-prod/images/AI_languages_blog_social.max-600x600.format-webp.webp">We’re moving beyond traditional text translation to build models that understand the world’s rich, living languages exactly as they are expressed.

On open weights for everyone else, credit for a disputed mathematics proof, and control over autonomous weapons. Source

OpenAI confirms weeks of AI safety talks with Anthropic and Google DeepMind, as Trump's team dismisses safety concerns and pushes to keep pace with China.

Profound announced on September 15, 2026, that it has raised a $180 million Series D at a $1.8 billion valuation, co-led by Sequoia Capital and Kleiner Perkins. The round arrives less than seven months after the AI marketing company closed its $96 million Series C. Existing investors Lightspeed Venture Partners, Khosla Ventures, Saga Ventures, Evantic, and South Park Commons all participated in the round. Profound said the new funding will expand its capabilities and support further research…

IBM Research on September 15, 2026, published an account of work proving unconditional theoretical separations between shallow quantum circuits and large language models: one functional problem and one sampling problem for which shallow quantum circuits hold a provable advantage over LLMs. The post's byline lists Srinivasan Arunachalam, Arkopal Dutt, Hari Krovi, Rik Sengupta, and Ryan Mandelbaum. It describes the paper "Separating quantum circuits from classical LLMs," by Arunachalam, Dutt…

A deepdive into how Codex has “taken over” OpenAI, how the frontier lab builds its agentic software factory, and the engineering challenges of one billion users. Details from inside OpenAI

Explore Meta One’s pricing and features for creators and businesses, including scheduling, analytics, AI tools and paid discovery.

How moving randomness outside the computation graph turns noisy gradient estimators into low-variance, differentiable ones The post Reparameterization Tricks: Variance Reduction by Smarter Gradients appeared first on Towards Data Science .

Too often, delivery robots just don't understand the rules of the road (or sidewalk).

Article URL: https://www.formas.ai/cartesian Comments URL: https://news.ycombinator.com/item?id=49713999 Points: 64 # Comments: 60

Meta introduced Meta One on September 15, 2026, a subscription service that adds paid tiers across Instagram, Facebook, WhatsApp, and Meta AI, with plans for individuals, creators, and businesses. Meta said the service launches with more than 50 features and 15 million subscriptions and trials to date. The company said the core experience across its apps and Meta AI will stay free, with subscriptions unlocking more specialized capabilities and expanded AI usage beyond the free tier. Plans are…

Despite the recent deluge of videos of humanoid robots doing backflips and kung-fu, the actual pace of progress towards humanoids that can do economically viable jobs at scale has at times seemed a bit plodding. There are some good reasons for this : a humanoid robot needs to be powerful enough to do useful work, safe enough for humans to walk past, and have enough battery life to limit how much of the day it’s hooked to a charger. But all three of those necessary features actively work against

Crusoe on September 15, 2026, announced a multi-year partnership with Perplexity under which Perplexity will run its full model lifecycle on Crusoe Cloud, training frontier models on dedicated NVIDIA GB300 NVL72 clusters and serving them in production through Crusoe's Managed Inference service. The announcement, datelined San Francisco, also commits Crusoe to adopting Perplexity Enterprise Pro and Max for its 1,800 employees. According to the release, the deployment is meant to equip staff with…

Anthropic announced the beta release of Salesforce in Claude on September 15, 2026, a plugin built with Salesforce that brings a seller's accounts, opportunities, and pipeline into Claude under their existing Salesforce permissions. Anthropic said the plugin includes 37 skills for the work account executives do daily, including account research, call prep, pipeline review, and CRM updates. According to Anthropic, sellers often spend hours of their day on meeting prep or follow-up work from…

Article URL: https://heymeraki.substack.com/p/aie_10-building-it Comments URL: https://news.ycombinator.com/item?id=49713671 Points: 8 # Comments: 2

Shortly after launching its new do-everything AI assistant Muse, Meta's launching subscription bundles that pair its standalone app subscriptions with extra AI usage. Some of the new Meta One bundles were in testing earlier this year, but are now available globally starting today, with several tiers for individual users, creators, and businesses. Meta says the […]