AI News Today

A new hope for treatment. ScienceAlert stories are written, fact-checked, and edited by humans, never generated by AI. Don't miss a story, subscribe here.

Keybridge lets an agent or a script run an npm publish that ends in a Touch ID tap instead of dying with EOTP. It automates npm's WebAuthn hand-off (the browser hop, the one-time-token relay, the retry) in a windowless WKWebView, and answers the ceremony from a Secure Enclave key that only signs after a live Touch ID. macOS, fully open source. Comments URL: https://news.ycombinator.com/item?id=48929255 Points: 2 # Comments: 0

<p><strong>Pre-dispatch authorization for AI sub-agents</strong> means checking a child spawn's grant envelope (its role, tools, path scope and token budget) against the parent's policy in the last plaintext moment before handoff, not by reading a trace afterwards. When the orchestrator encrypts the handoff, the after-view goes blind. The before-view does not, because it sits earlier on the timeline.</p> <p>On July 14, "Codex starts encrypting sub-agent prompts" hit <a href="https://news.ycombin

<pre style='white-space:pre-wrap;width:81ex'>improve: speed up Control UI browser tests (#108504) * test(ui): accelerate browser timing coverage * test(ui): stabilize Mantis elapsed-time proof * test(ui): keep Claude viewport proof isolated</pre>

<p>Every agent-security vendor tells you what they block. Nobody tells you what they miss.</p> <p>That gap is the whole problem. "We stop prompt injection" is a claim you cannot check. You cannot run it, and you cannot tell it apart from the next company saying the same sentence. So security engineers do the rational thing and discard all of it.</p> <p>I published the opposite. It is called the <strong>ARE Incident Database</strong>, and it is public: <a href="https://aredb.org" rel="noopener no

<p> AI account executives that demo and close B2B deals </p> <p> <a href="https://www.producthunt.com/products/river-9?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/1197712?app_id=339">Link</a> </p>

Article URL: https://twitter.com/Akashi203/status/2077552491567157733 Comments URL: https://news.ycombinator.com/item?id=48929104 Points: 2 # Comments: 0

<p>If you are like me and use AI often to help with complicated tasks and own a Flipper Zero, you might want a Super Tiny Language Model (STLM). </p> <h2> Why this is useful </h2> <p>Until now, there have been no language models small enough to run on a small device, such as a Flipper Zero. Flippermind relies on lightweight Python3 installs and a local connection to Qwen2.5-0.5B. Although this is configured for Flipper Zero, you can run it on other small machines like a Raspberry Pi.</p> <h2> Ch

<p>Sleep is the cornerstone of health, yet millions suffer from undiagnosed sleep apnea. If you've ever wondered about the quality of your rest but felt uneasy about uploading hours of private bedroom audio to the cloud, you're in the right place. In this tutorial, we are building a privacy-first <strong>Sleep Snoring Monitoring System</strong> using <strong>Faster-Whisper</strong> and <strong>Voice Activity Detection (VAD)</strong>.</p> <p>By leveraging <strong>local AI deployment</strong> and

<p><strong>Tool:</strong> <a href="https://tools.simonwillison.net/grok-mermaid">Mermaid to Unicode box art (grok-mermaid)</a></p> <p>While <a href="https://simonwillison.net/2026/Jul/15/grok-build/">exploring the codebase</a> for the newly open-sourced Grok CLI coding agent I came across <a href="https://github.com/xai-org/grok-build/blob/b189869b7755d2b482969acf6c92da3ecfeffd36/crates/codegen/xai-grok-markdown/src/mermaid.rs">xai-grok-markdown/src/mermaid.rs</a>, a "self-contained terminal ren

Article URL: https://www.youtube.com/watch?v=iVkgD4phOHQ Comments URL: https://news.ycombinator.com/item?id=48929052 Points: 1 # Comments: 0

Article URL: https://github.com/JustMichael-80/MasterVault Comments URL: https://news.ycombinator.com/item?id=48928934 Points: 1 # Comments: 0

Thinking Machines' first open weights model is a 975 billion parameter alternative to Chinese LLMs

Article URL: https://www.aicss.dev Comments URL: https://news.ycombinator.com/item?id=48928924 Points: 1 # Comments: 0

Kevin here, indie dev (lorisLabs.) Thorttle started as a menu bar meter for Claude code usage, and grow into a cockpit ! you have the full claude meter as expected with no telemetry, no network. and that in the free version. Pro is where it got interesting : Cockpit have every claude code project on your Mac in one window. Each embedded terminal using SwiftTerm, auto hibernate to keep a 16gb Mac from getting out of memory. Remote Control that hand a session off to any linux server (I use proxmox

<pre style='white-space:pre-wrap;width:81ex'>fix(line): use precise control-command check for group mention bypass (#107230) * fix(line): use precise control-command check for group mention bypass Group requireMention bypass is gated on `command.hasControlCommand`, but LINE fed it the broad `shouldComputeCommandAuthorized` detector, which is true for any inline "/x"/"!x" token. An allowlisted member's plain message like "cd /home" would satisfy the bypass and reach the agent even though the bot

<p>When the PM talks about new laws applying to the ‘next generation of large-scale datacentres’ what does he mean?</p><ul><li><p><a href="https://www.theguardian.com/australia-news/2026/jul/16/albaneses-ai-blueprint-sparks-calls-for-datacentre-moratorium-until-new-regulations-in-place">Albanese’s AI blueprint sparks calls for datacentre moratorium until new regulations in place</a></p></li></ul><p>Expectations were high as the prime minister took the stage at the University of Sydney on Wednesd

<pre style='white-space:pre-wrap;width:81ex'>fix(agents): prevent subagent registry resurrection (#105793) * fix(agents): retire legacy subagent registry JSON Co-authored-by: morluto <76467478+morluto@users.noreply.github.com> * fix(agents): keep migration decision internal Co-authored-by: morluto <76467478+morluto@users.noreply.github.com> * style(agents): format migration decision type Co-authored-by: morluto <76467478+morluto@users.noreply.github.com> --------- Co-authored-by: Peter Steinberg

The ISR Observability page now shows Observability Plus subscribers a write utilization metric, helping identify routes that regenerate often but receive few requests. Write utilization is the ratio of cached requests to ISR writes. For routes with low write utilization, consider increasing the revalidation interval or switching to on-demand revalidation to reduce costs. You can also compute write utilization using the Vercel CLI, or use agents with the cdn-caching skill to investigate. Learn mo

Kimi K3 from Moonshot AI is now available on AI Gateway. K3 is an open-source model with a 1M-token context window and native visual understanding, accepting text, image, and video inputs. Built for long-horizon software engineering, knowledge work, and deep reasoning, K3 is especially strong where code meets visual and spatial reasoning, which suits frontend, game development, and CAD workflows. Thinking mode is always on. To use Kimi K3, set model to moonshotai/kimi-k3 in the AI SDK : AI Gatew

Cars24 uses OpenAI-powered voice and chat agents to handle 1M+ monthly conversation minutes, recover 12% of lost leads, and bring agentic workflows to teams across the company.

Incremental video search requires high-quality ranking after each keystroke, where intent is often underspecified (e.g., 1–3 character prefixes). We present a personalization system for Apple TV search that combines complementary semantic and collaborative signals at ranking time. Our approach learns two item embedding spaces: (i) a text-based multilingual encoder (TextEmb) fine-tuned on co-engagement triplets via contrastive learning, and (ii) an ID-based collaborative embedding model (IdEmb) t

Can a large language model (LLM) improve at code generation using only its own raw outputs, without a verifier, a teacher model, or reinforcement learning? We answer in the affirmative with simple self-distillation (SSD): sample solutions from the model with certain temperature and truncation configurations, then fine-tune on those samples with standard supervised fine-tuning. SSD improves Qwen3-30B-Instruct from 42.4% to 55.3% pass@1 on LiveCodeBench v6, with gains concentrating on harder probl

Microsoft is looking to sell its in-house AI models as more efficient and cost-effective than its competitors' models.

Article URL: https://lore.kernel.org/linux-media/CAHk-=wi4zC+Ze8e+p3tMv8TtG_80KzsZ1syL9anBtmEh5Z40vg@mail.gmail.com/ Comments URL: https://news.ycombinator.com/item?id=48928697 Points: 11 # Comments: 2

Article URL: https://lhv.ai/ Comments URL: https://news.ycombinator.com/item?id=48928679 Points: 1 # Comments: 0

Article URL: https://www.kqed.org/news/12090173/too-old-for-silicon-valley-think-again-ai-is-changing-the-math Comments URL: https://news.ycombinator.com/item?id=48928462 Points: 3 # Comments: 0

Location:Plymouth, MA (Greater Boston Area) Remote:Yes Willing to relocate:No(I have 3 kids under 6 currently!), but always willing to travel! Technologies: Python, Ubuntu, Docker, Machine-learning, Unity engine, VS, SQL Oracle, C++, C#, AWS, OVH, Django. Résumé/CV: https://www.linkedin.com/in/robert-mcdermott-56693864/ Email: robert.mcd@live.com Currently a student at BHCC as a Computer Science Major with a focus in Software Engineering and AI. I just want to get my foot in the door somewhere w

Article URL: https://coroot.com/blog/hard-part-of-ai-root-cause-analysis-is-no-longer-the-model/ Comments URL: https://news.ycombinator.com/item?id=48928324 Points: 4 # Comments: 1

Article URL: https://undark.org/2026/07/15/medicare-prior-authorization-ai/ Comments URL: https://news.ycombinator.com/item?id=48928319 Points: 3 # Comments: 0

General-purpose robots and autonomous machines are moving from research labs to real-world mass-market deployment, creating demand for compact, power-efficient AI supercomputers capable of running foundation models at the edge. To meet that need, NVIDIA today introduced the T3000 and T2000, new modules based on the NVIDIA Thor architecture that enable mass-market robotics and edge AI […]

Developers building video analytics applications across large spaces must track the same object as it moves between camera views. Single-camera 2D tracking...

Article URL: https://eaon.dev Comments URL: https://news.ycombinator.com/item?id=48928305 Points: 2 # Comments: 1

Article URL: https://www.noahpinion.blog/p/why-i-didnt-sign-the-we-must-act Comments URL: https://news.ycombinator.com/item?id=48928055 Points: 6 # Comments: 0

One-two punch offers a glimpse of how low-precision AI can complement high-precision simulations

h5i-python is the Python SDK for the h5i orchestra engine. This SDK lets you define and execute multi-agent coding workflows across Claude Code, Codex, and other runtimes as ordinary Python programs. Comments URL: https://news.ycombinator.com/item?id=48927933 Points: 2 # Comments: 0

Across 101 enterprises, agent orchestration is consolidating onto model-provider platforms — Anthropic’s Claude leads by a wide margin — chosen for the gravity of the underlying model and judged on reliable multi-step execution. But the ambition runs well ahead of the reality: most deployed “agents” are still chatbot wrappers, the control plane enterprises expect is deliberately hybrid to avoid lock-in, and real-time fiscal control over token burn remains the exception. This wave of VentureBeat

Just wondering whether others have the same impression: are the articles better on the second page of HN? The front page is basically AI and associated slopationalism while the second (and further) pages are more down to earth and worth investigating. Comments URL: https://news.ycombinator.com/item?id=48927922 Points: 6 # Comments: 2

Article URL: https://blog.greg.technology/2026/07/14/llm-networking-with-mikrotik.html Comments URL: https://news.ycombinator.com/item?id=48927915 Points: 1 # Comments: 0

Article URL: https://a16y.ai Comments URL: https://news.ycombinator.com/item?id=48927838 Points: 4 # Comments: 0