AI News Today

Software supply chain security was hard enough. Then AI joined the build pipeline. For five years, "software supply chain security" meant one question: what's in your code? Which open-source packages, which versions, which transitive dependencies three layers deep that nobody chose on purpose? SolarWinds, Log4Shell, and XZ Utils all taught the same lesson: the risk lives less in the code a

With the rapid progress of AI capabilities and the move to agentic systems, organizations are expanding their use cases as the technology continues to grow. That constant evolution also introduces risk, leaving IT leaders to wonder which investments will prove valuable even six months into the future. Returning to the foundational elements of AI architecture—the…

Chinese startup Deepseek is building its own AI chip, Reuters reports. The article Deepseek is designing its own AI chip appeared first on The Decoder .

Top robotics researchers and founders explain how robot autonomy is evolving.

Top robotics researchers and founders explain how robot autonomy is evolving.

Do any renowned OSS projects (e.g. OS, compilers, databases, libraries...) switched their review process and started allowing vibe-coded PR to be merged? A few months ago, there was this AI generated PR to the OCaml project ^1 that famously got rejected. Has the stance changed lately with increasingly better LLM models? --- ^1 https://github.com/ocaml/ocaml/pull/14369 Comments URL: https://news.ycombinator.com/item?id=48816039 Points: 1 # Comments: 1

OpenAI, Anthropic, and major cloud providers are racing to outbid each other with free compute credits to pull startups into their ecosystems. Some individual offers top $3 million. At Y Combinator alone, OpenAI and Anthropic could hand out up to $800 million in credits per year combined. The discount war comes at a time when both companies need to improve their margins ahead of upcoming IPOs. The article OpenAI and Anthropic are giving away millions in computing power to attract startups appear

Not sure this will have any effect, but I support the effort: According to Google’s legal filing, Outsider Enterprise operates through Telegram. The group offers phishing-as-a-service to individuals who may not be technically savvy enough to set up fraudulent websites and text campaigns on their own. In its Telegram channels, Outsider Enterprise reportedly provided instructions on how to use Google’s Gemini AI to create websites that imitate those of Google, YouTube, and government agencies such

Article URL: https://moneymagician.eu Comments URL: https://news.ycombinator.com/item?id=48815980 Points: 1 # Comments: 0

Apollo chief economist Torsten Slok sees no AI-driven margin gains outside tech. In regulated industries like healthcare, banking, or pharma, process overhauls and privacy rules could delay productivity boosts by years. If that takes five years instead of five months, many AI stocks face a painful repricing. The article Apollo economist warns AI profit gains outside tech could take "well beyond" what Wall Street expects appeared first on The Decoder .

Article URL: https://www.ft.com/content/304cdbb9-c161-4e40-af4d-6df1d2ff5363 Comments URL: https://news.ycombinator.com/item?id=48815867 Points: 2 # Comments: 0

Article URL: https://theconversation.com/abc-will-trial-using-ai-for-journalism-what-are-the-risks-and-benefits-286841 Comments URL: https://news.ycombinator.com/item?id=48815818 Points: 1 # Comments: 0

Article URL: https://prevail.sh/ Comments URL: https://news.ycombinator.com/item?id=48815784 Points: 2 # Comments: 0

Article URL: https://brianni.co/blog/prove-were-not-lying Comments URL: https://news.ycombinator.com/item?id=48815739 Points: 1 # Comments: 0

Norwegian striker Erling Haaland isn’t just a footballer anymore. He’s become an internet character perpetuated by fans and AI.

Are you tired of manually copying and pasting data between AI tools? Looking for an agentic AI tool that will pass off multi-step processes to other tools without constant input from you? In this article, you'll discover how to get started using Manus to build agentic workflows that delegate complex tasks to other tools with […] The post Building Agentic Workflows with Manus appeared first on Social Media Examiner .

It's getting hot in here (relatively). ScienceAlert stories are written, fact-checked, and edited by humans, never generated by AI. Don't miss a story, subscribe here.

L’Oreal is using AI to shorten product development timelines and identify new uses for ingredients already present in its portfolio. The French cosmetics group has applied AI in its laboratories for the past four years, with Fabrice Megarbane, president of L’Oreal’s consumer products division, telling Reuters that the technology helps the company predict how molecules […] The post L’Oreal, Mondelez, and Nestle use AI to speed product development appeared first on AI News .

<p> The data layer for AI agents </p> <p> <a href="https://www.producthunt.com/products/zoodata?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/1190090?app_id=339">Link</a> </p>

Article URL: https://www.bbc.com/news/articles/c1kyy8yrpxdo Comments URL: https://news.ycombinator.com/item?id=48815657 Points: 5 # Comments: 3

The lab will beam back data to train AI models to predict how proteins behind age-related diseases like Alzheimer’s and certain cancers behave.

Committee says Anthropic's brief US export ban shows UK can't even trust its allies to keep the AI lights on

<p>Every WooCommerce integration project we take on now ends with the same question from the client: "can the AI just check the store for me?" What were last month's sales, which SKUs are out of stock, has order 48821 shipped. Reasonable questions. The tempting answer is to hand an agent full API access and let it figure things out. That is also how you end up with an LLM editing prices at 2am because it "helpfully" corrected what it thought was a typo.</p> <p>So we built the boring version inst

<p>Andrew Charlton says artificial intelligence ‘cheating, deceiving, going their own way’ – and time to get ahead of it is during testing </p><ul><li><p><a href="https://www.theguardian.com/australia-news/live/2026/jul/07/australia-news-live-penny-wong-china-missile-test-pacific-anthony-albanese-solomon-islands-defence-royal-commission-antisemitism-meta-ntwnfb">Follow our Australia news live blog for latest updates</a></p></li><li><p>Get our <a href="https://www.theguardian.com/email-newsletter

<p>Ask a team "which models is your application allowed to call, and under what conditions?" and the honest answer is usually <em>"let me check the code."</em> The rules — which models are approved, which tools an agent may invoke, what happens when a request is too large or comes from the wrong region — are scattered across <code>if</code> statements in a dozen services. No one can review them in one place, no one can test a change safely, and no one can say what a rule <em>would have done</em>

As of July 7, the iRobot Roomba 105X robot vacuum and mop is discounted to $219.99 at Amazon. That's 53% off its list price of $469.99.

<pre style='white-space:pre-wrap;width:81ex'>refactor(agent-core): localize exec timeout helper (#101523)</pre>

Article URL: https://github.com/lambda-alpha-labs/Graphenium Comments URL: https://news.ycombinator.com/item?id=48815458 Points: 1 # Comments: 0

Three new reports show how AI power is concentrating, and why participation has to include the right to refuse. Source

Article URL: https://reykur.io/blog/ai-coding-assistant-shipping-secrets/ Comments URL: https://news.ycombinator.com/item?id=48815447 Points: 3 # Comments: 0

<p>Fallbacks make AI products feel safer. They can also make the bill harder to explain.</p> <p>A typical production setup no longer sends one request to one model and stops. It may try a preferred model, retry after a timeout, move to a backup route, shorten context, call a smaller model for classification, then call a larger model for the final answer.</p> <p>That is often the right engineering choice. The problem starts when the team only watches the first model name or the final HTTP status.

Article URL: https://pixelglass.co/ Comments URL: https://news.ycombinator.com/item?id=48815350 Points: 2 # Comments: 0

Paste one line of code. FaultFixer detects every front-end and back-end error, diagnoses the root cause with AI, and hands it straight to your AI coding tool. Comments URL: https://news.ycombinator.com/item?id=48815345 Points: 1 # Comments: 0

<p>For most of my career I had backups I had never restored. The backup job ran every night. The green checkmark appeared every morning. The storage bucket filled up with files that had the right names and roughly the right sizes. I looked at all of that and told myself the data was safe. What I never did was take one of those files, restore it into a clean database, and confirm that the result was a working copy of production. I trusted the checkmark, and the checkmark was measuring the wrong t

<p><em>Part of <a href="https://github.com/gabriele-mastrapasqua/qwen3-tts" rel="noopener noreferrer">qwen3-tts</a> — a pure C inference engine for Qwen3-TTS.</em></p> <h2> TL;DR </h2> <p>The engine is pure C and <strong>CPU by default</strong>. We added two <strong>opt-in</strong> GPU backends that leave the CPU path untouched:</p> <ul> <li> <strong>Apple Metal</strong> (<code>make metal</code>) — 0.6B model <strong>RTF 1.5 → 0.60 on an M1</strong>, <strong>0.36 on an M2 Pro</strong>; streaming

Article URL: https://arxiv.org/abs/2602.15654 Comments URL: https://news.ycombinator.com/item?id=48815301 Points: 1 # Comments: 0

<pre style='white-space:pre-wrap;width:81ex'>fix(agent): preserve explicit recipient sessions (#101507) * fix(agent): honor recipient session routing Co-authored-by: vincentkoc <25068+vincentkoc@users.noreply.github.com> Co-authored-by: pingfanfan <pingfan.work@gmail.com> * fix(agent): preserve canonical recipient routes * fix(agent): require exact recipient routes Co-authored-by: vincentkoc <25068+vincentkoc@users.noreply.github.com> Co-authored-by: pingfanfan <pingfan.work@gmail.com> * fix(age

<p>Large-scale datacentre projects around the world are being challenged or cancelled, as infrastructure’s energy demands ramp up</p><p>Datacentre planning proposals face all kinds of hurdles, from securing energy supply to high construction costs. But the 2,000 acre Prince William Digital Gateway site in the US state of Virginia had another problem: its proximity to a Civil War battlefield.</p><p>“If the development is allowed to proceed, the solemn nature of this historic site would become mar

Forterra has deployed more than 100