📰 Alle News

← Command Center
Arm Neoverse CSS N4: Baukasten für Server-CPUs mit 128 Kernen und PCIe 7.0

Neben neuen Smartphone-Kernen wird Arm in Kürze auch mit neuen Kernen nach Baukastenprinzip für Profichips nachlegen. Zuerst kommen dabei die neuen Neoverse N4 als Compute Subsystem (CSS), ein von Arm...

heise+ | Vorstellung Dacia Spring: Mit Wiedererkennungswert

Dacia bringt den neuen Spring auf den Markt. Der elektrische Kleinwagen ist größer, moderner und soll trotz technischer Upgrades bezahlbar bleiben.

Einstein hat schon wieder recht: Forscher bestätigen 119 Jahre alte Idee für Quantensysteme

Präzisionsmessungen eines internationalen Forschungsteams an ultrakalten Quantengasen liefern Ergebnisse, die helfen könnten, eines der letzten großen Rätsel deweiterlesen auf t3n.de

Intel belichtet eine Million Wafer mit ASMLs 350-Millionen-Euro-Systemen

Intel setzt als erster Chipfertiger ASMLs neueste Lithografie-Systeme ein. Parallel läuft die Entwicklung größerer Masken.

Ein Start-up entfernt Verweigerung aus KI-Modellen und verdient daran

Das Start-up Abliteration AI entfernt Verweigerungsmechanismen aus KI-Modellen. Das hilft Security-Teams, schafft aber auch neue Missbrauchsrisiken.

Navier-Stokes – Tristan Buckmaster [pdf]

Article URL: https://cims.nyu.edu/~tristanb/statement.pdf Comments URL: https://news.ycombinator.com/item?id=49605915 Points: 108 # Comments: 19

Uber in Afrika in Schwierigkeiten

Der Fahrdienstvermittler hat in Nigeria und Uganda den Geschäftsbetrieb mit sofortiger Wirkung eingestellt. In Nigeria ermittelt die Wettbewerbsbehörde.

My Feed, My Way

Article URL: https://www.pm.gov.au/media/my-feed-my-way Comments URL: https://news.ycombinator.com/item?id=49605782 Points: 105 # Comments: 70

Mistral raises €3B to make sovereign, open-weight AI the technology frontier

Article URL: https://mistral.ai/news/mistral-makes-sovereign-open-weight-ai-to-frontier/ Comments URL: https://news.ycombinator.com/item?id=49605767 Points: 127 # Comments: 54

heise+ | Produktionsreife Container-Plattform auf macOS im Test

Apples Container-Engine für macOS hat das erste stabile Release erreicht. Drei Open-Source-GUIs ergänzen sie um weitere Funktionen.

Satelliteninternet für Handys: Telekom spricht mit Vodafone, Orange, Telefónica

Die EU will Starlink-Alternativen fördern und ihnen Spektrum für Satelliteninternet auf Smartphones reservieren. Darum könnte ein mächtiges Konsortium bieten.

We have a year to fix security everywhere

Article URL: https://jyn.dev/a-year-to-fix-security/ Comments URL: https://news.ycombinator.com/item?id=49605691 Points: 111 # Comments: 71

Huawei proudly shows off an entirely un-American chip

Resistance is futile

CSS for Mobile 2: ARM will weiterhin ohne NPU auskommen

ARM hat eine neue Generation an CPU-Kernen für Smartphones vorgestellt. Die sieht Matrixeinheiten für KI bereits im Einstiegssegment vor. (Prozessor, Smartphone)

Mali G2-Ultra NX: KI-Rendering ist auch bei ARM angekommen

Bessere Grafik, höhere Frameraten bei gleichzeitig niedrigerer Leistungsaufnahme verspricht ARMs neue Mali-GPU-Generation dank KI-Kernen. (Grafikkarten, KI)

Arm CSS for Mobile 2: C2-Ultra macht die CPU zum Orchestrator für KI‑Agenten

Das Arm CSS for Mobile 2 soll das Arm-Ökosystem auf das Zeitalter der KI-Agenten vorbereiten. Die Plattform künftiger Arm-Prozessoren besteht im Zentrum aus der neuen C2-Ultra-CPU mit doppelt so viele...

Arm Mali G2 NX: GPUs rendern nur noch ein Achtel der Grafik, der Rest ist KI

Arm stattet die GPUs der Mali-G2-NX-Generation erstmals mit dedizierten Neural-Beschleunigern aus, die Aufgaben wie Upscaling, Frame Generation und Denoising übernehmen. Damit lassen sich höhere Frame...

220 million traveler records exposed in Vietnam-linked APIS leak

Exclusive: An exposed Advance Passenger Information System (APIS) database held 220 million passenger and crew records containing names, passport numbers, dates of birth, nationalities, and flight det...

Xiaomi’s wide foldable promises more power than Samsung’s

Xiaomi is the latest manufacturer to release a short, wide foldable phone, in the same week Apple is expected to do the same. I got to try the 18 Fold behind closed doors at IFA last week, and while i...

How well do agents use test/verification techniques?

Article URL: https://danluu.com/agentic-testing/ Comments URL: https://news.ycombinator.com/item?id=49605246 Points: 101 # Comments: 30

Eric Wu’s newest company, out of stealth since May, is going after construction’s labor crunch

Eric Wu, who built and ran Opendoor before stepping away in 2022, has had his new company, NavigateAI, out of stealth since May — building AI copilots that give construction workers real-time, hands-f...

There's a new "Google Jail" for independent wikis

Article URL: https://weirdgloop.org/blog/google-jail Comments URL: https://news.ycombinator.com/item?id=49604870 Points: 106 # Comments: 40

Jellyfin 12.0

Article URL: https://jellyfin.org/posts/jellyfin-release-12.0/ Comments URL: https://news.ycombinator.com/item?id=49604861 Points: 120 # Comments: 48

I've factored the RSA keys of a Certificate Authority from the 90s

Article URL: https://mcpherrin.ca/2026/09/07/rsa.html Comments URL: https://news.ycombinator.com/item?id=49604637 Points: 115 # Comments: 23

The Wispr Flow Notetaker is here - and it's got all the context with none of the bots (Sponsor)

We've all accepted that notetakers get things wrong, and all the pain that comes with: tedious cleanup, garbled customer quotes, confusing action items.But what if your notetaker got it right the firs...

Anthropic signed $517bn in compute agreements in past 11 months (2 minute read)

Anthropic has secured $517 billion in compute capacity leases over the past 11 months, amounting to 14.8GW, primarily with Google and AWS. This expansion includes large deals with cloud providers like...

OpenAI prepares managed agents for DevDay 2026 (3 minute read)

OpenAI plans to introduce Managed Agents at DevDay 2026, following a model similar to Anthropic's offerings, targeting businesses and developers. The new agents aim to integrate advanced models with s...

Everyone is lying to your CRM (Sponsor)

Not on purpose! But stages are always optimistic and close dates go stale. People are busy... Lightfield is the agentic CRM that builds the record from the conversations themselves, capturing every em...

ChatGPT May Soon Learn Your Writing Style From Your Slack or Gmail (2 minute read)

OpenAI appears to be implementing a new feature that helps ChatGPT write in users' voices by referencing examples from their connected apps. Some users have had the feature for weeks, but the company ...

The Two MMLU Scores: What a Benchmark Name Does Not Fix (23 minute read)

Two builds with the same provider, model family, metric identifier, unit, and benchmark name can score differently. The difference is the runners, graders, and dataset splits. The shared mmlu label id...

Automatically detecting AI text in my browser (5 minute read)

Automated AI text detection is currently an underserved niche. Pangram does an excellent job, but it is still more a tool to confirm suspicion. This developer wanted a tool that runs in the background...

ByteDance Joins AI Elite in Race to Perfect World Models (5 minute read)

ByteDance is preparing to launch a real-time spatial video AI model. The model will be able to create interactive virtual worlds. It will also be able to generate virtual worlds that respond to Pico V...

XPeng starts IRON humanoid robot production as Tesla Optimus stalls (5 minute read)

XPeng has started production for its IRON robot. Its production line runs with more than 80% of its core processes automated. The company's robot went viral last year as its walk was so smooth that pe...

Early Data Indicates an AI-Generated Drug Could Slow Aging (2 minute read)

Rentosertib is a drug created with the assistance of AI. The AI originally indicated that the drug could help treat patients suffering from a chronic lung disease, but data from clinical trials show i...

Prompt Injection Through Tool Output (8 minute read)

Prompt injections hidden in tool results can evade conventional safeguards because input and action screens inspect separate moments of an agent loop. The proposed signal is a “precedent gap,” where a...

TPU Inference Externalization Full Steam Ahead (35 minute read)

Google's TPUv7 Ironwood delivers up to 50% better performance per dollar compared to Nvidia's B200/B300 chips. Ironwood is the first generation in which Google is competing for others' inference workl...

Your AI agents need guardrails and boundaries (Sponsor)

What are your AI agents doing? What do they have access to? Teleport gives you answers and helps you set boundaries to contain and control them. Monitor and audit agent behavior to better understand t...

The Year Finding and Exploiting Bugs Became Cheap, and What to Do About It (13 minute read)

Over the next few years, offensive capability will become abundant. More people will be able to inspect every valuable project and exploit vulnerabilities. Handling critical bugs in a timely manner wi...

How well do agents use test/verification techniques? (68 minute read)

It's easier than ever to hit a particular quality bar by having coding agents use effective test techniques. However, software quality seems to be getting worse. This indicates that whatever defaults ...

Four Questions About AGI (28 minute read)

Artificial General Intelligence (AGI), as the term is commonly used, points to nothing really tangible or concrete and points us at the wrong thing. Recent announcements about AGI are just companies s...

Your AI deserves better than garbled transcripts (Sponsor)

If your AI is writing nonsensical followups and misattributing customer quotes, you might want to double-check your transcript... or use the Wispr Flow Notetaker and get meeting transcripts you can tr...

Google Accelerator Agents for TPU Development (GitHub Repo)

Google's Accelerator Agents use Gemini to help developers migrate PyTorch workloads to JAX and optimize custom kernels for Google Cloud TPUs. The toolkit includes MaxCode for model conversion and MaxK...

Qwen-Drive (GitHub Repo)

Qwen-Drive is a project that aims to create a Vision-Language Foundation model for autonomous driving. The project uses a staged training strategy that integrates perception, language, and planning ob...

hip-agent: a harness that fits in the prompt (5 minute read)

hip-agent is a small agent harness designed for agents. The configuration is environment variables, actions are shell commands, and a subagent is a child process. The rest is handled by existing proto...

The Education of a Doomer (9 minute read)

Automation has been good, but AGI may make humans economically useless, which will have many consequences. It's clear that AI capabilities are growing far faster than our ability to control or underst...

The Chasm: The Shape of Unfinished AI Codebases (6 minute read)

AI codebases exhibit a unique, unpredictable pattern where programs appear polished but often hide deep, hidden issues leading to significant failures outside controlled demos. Unlike human-authored p...

De-Brainrot Vacations (6 minute read)

The monotony of work can make it hard to keep a clear head. Spend your vacations doing something different for your brain. Slow down and double down on new habits you want to grow. Spend time in natur...

Machines that think: embodied intelligence (10 minute read)

Vision and language models are making strides, but embodied AI in robotics struggles due to sparse, costly training data for manipulation tasks. Robotics excels where tasks are fixed, yet general-purp...

What We Can Learn from Claude's Fable 5.1 System Prompt (14 minute read)

Comparing how system prompts develop over time demonstrates how model quirks and product design are moving targets.

Simple is not small (11 minute read)

Writing simple programs requires a good mental model of the program as well as good taste.

← Neuere Seite 101 Ältere →