Tech News

Nature Machine Intelligence

Các nhà khoa học phát hiện ra rằng việc dạy máy tính suy nghĩ đôi khi giống như việc thả bi lăn vào một mê cung ngoằn ngoèo đến mức dù bạn có cẩn thận đến đâu, chỉ cần lệch một chút xíu là viên bi sẽ đi đến một chỗ hoàn toàn khác.

IEEE Spectrum

Các nhà khoa học dùng máy ảnh đặc biệt và chương trình máy tính để chụp ảnh mặt trời khi mặt trăng che khuất nó, giúp hiểu rõ hơn về lớp khí quyển ngoài cùng của mặt trời.

Google Research

Health & Bioscience

OpenAI

OpenAI introduces its National Community Compact and Compute Campus Policy Blueprint to protect local resources, create jobs, and guide responsible AI infrastructure development.

OpenAI

OpenAI built an autonomous pipeline that found, validated, and generated patches for 948 local privilege-escalation vulnerabilities in the Linux kernel.

Hugging Face

OpenAI

DevRev uses GPT-6.1 Sol to power Computer and Operator, balancing reasoning, speed, and cost for enterprise service.

Google Research

Data Mining & Modeling

Đây là một mô hình trí tuệ nhân tạo mã nguồn mở giúp máy tính có thể nói chuyện bằng giọng nói nghe tự nhiên như con người.

214 stars · 55.6k total

Đây là một công cụ giúp các chương trình máy tính thông minh tự động mở và đọc nội dung trên các trang web như con người.

181 stars · 45.7k total

Đây là một bộ hướng dẫn giúp trợ lý trí tuệ nhân tạo hiểu cách sử dụng các dịch vụ của Google giống như một chuyên gia.

199 stars · 22.2k total

The protection of mature, multi-stage Face Recognition Systems is crucial to system providers. This technical report targets the threat model most relevant to industry practitioners: although model training is performed in-house, training data is sourced from possibly-compromised

A central challenge in AI is specifying objectives that align with human preferences. While existing theoretical results elucidate how misspecification impacts performance, they typically do not distinguish between merely suboptimal and catastrophic outcomes. To enable rigorous a

Tiered inference distributes large language model computation across a hierarchy of devices, from on-device and edge hardware to the datacenter, and is increasingly offered as the route to cheaper, lower-latency, and more private serving. Most such systems orchestrate frozen mode

We use sparse autoencoders (SAEs) to "brain scan" large language models, mapping internal activations to interpretable concepts and enabling targeted "steering" of specific features. We show that prompting models to adopt specific preferences activates many unintended concepts, c

Security-by-design emphasises integrating security into software design and development. Security pattern catalogues operationalise this principle by prescribing reusable architectural solutions to recurring security problems. However, recovering their implementations from source

Deep learning over relational databases has led to increasingly effective prediction models, motivating a growing need to explain their behavior. We study explanation methods that assign importance scores to individual database attributes, focusing on a fundamental question: to w

Multi-agent LLM applications repeat most of their prompt tokens across requests, and serving engines reuse the cached state of shared prefixes. vLLM and SGLang reuse it at different granularities, and benchmarks of their default deployments report large performance gaps, usually

Researchers use large language models as agents in social simulations, where agents are prompted with observations and background information intended to shape their actions. Yet LLM agents have been shown to substantially change their behavior under meaning-preserving prompt var

The persistent volatility of construction material prices poses significant risks to cost estimation, budgeting, and project delivery, underscoring the urgent need for granular and scalable forecasting methods. This study develops a forecasting framework that leverages the Constr

Aviation authorities hold decades of accident reports whose narratives record how investigators reconstructed each accident. Computational work on these archives has mostly sorted reports into fixed taxonomies. In this work, we propose a hybrid framework in which a text-based cor

Looped transformers apply a shared block multiple times and have emerged as a parameter-efficient route to scaling compute in language models. However, at fixed FLOPs a looped model has strictly less capacity than a baseline transformer. We argue that compute, the number of seque

The full-attention layers of large language models (LLMs) make every generated token more expensive than the previous one, slowing models down on long-context tasks. Hybrid models reduce this cost by replacing most full-attention layers with linear-attention layers, but the first

Small LoRA adapters trained on Tajik news and books lowered Qwen3's Tajik text loss at every size we tried, from 0.6B to 8B, by 0.60 to 1.45 nats per token, but none of them improved Tajik reading comprehension where it can be measured. On Belebele the 1.7B and 8B adapters left a

Tool descriptions tell an agent what an action can do, but they do not necessarily reveal whether that action is healthy or suitable at the moment of use. Execution harnesses often possess this information through monitors, traces, and recent failures while the model choosing the

LLM-as-a-judge is a common substitute for costly human evaluation, and aggregating several judges is the standard remedy for any single judge's bias. Judge pools mix judges of very different reliability, while the human-labeled sets that validate them hold a few hundred items or

Large language model (LLM) coding assistants and agents construct a repository context before generating code. For security-sensitive code, this context decides whether the model sees the project's safeguards (e.g., authorization checks) or insecure code to imitate. However, cont

Enterprise teams that deploy large language model (LLM) agents inherit a maintenance problem that traditional testing does not address. The model is non-deterministic, but the larger risk is that most of what changes an agent's behavior (a retired model version, a revised policy,

Infrastructure security does not establish whether an LLM agent selected an allowed tool, grounded its answer in tool output, or followed a required workflow. AgentTrustCI is a Java/JUnit 5 framework that expresses these application-layer expectations as policy, grounding, and tr

Generative AI is currently fragmented, with separate subscriptions required for text generation, image synthesis, and code assistance, leading to high costs and workflow disruptions. AI Express is a SaaS platform that consolidates these capabilities into a unified pay-as-you-go s

Open-weight language models that fit on a single consumer GPU are becoming capable on a growing range of tasks, and they matter because they can run locally without sending the code to a hosted service. Vulnerability reproduction, turning a vulnerability description into an input

Những người tình nguyện trên bách khoa toàn thư mở đang tạo ra các công cụ và quy tắc để tìm và xóa các bài viết do trí tuệ nhân tạo viết ra mà không có sự cho phép.

MIT Associate Professor Jacob Andreas of the Department of Electrical Engineering and Computer Science [EECS] and MIT Associate Professor Brett McGuire of the Department of Chemistry have been selected as the winners of the 2026 Harold E. Edgerton Faculty Achievement Award. Estab

Jack Dennis, an influential MIT professor emeritus of computer science and engineering, died on March 14 at age 94. The original leader of the Computation Structures Group within the MIT Computer Science and Artificial Intelligence Laboratory (CSAIL), he pioneered the development

Training a large artificial intelligence model is expensive, not just in dollars, but in time, energy, and computational resources. Traditionally, obtaining a smaller, faster model either requires training a massive one first and then trimming it down, or training a small one fro

UK government partners with Google DeepMind to build a new AI-powered prototype aimed at faster housing decisions.

Google DeepMind and partners announce a $10M funding call for multi-agent safety research.

Health & Bioscience

For some, rules pose limitations. For others, like Microsoft Research Manager Mike Walker, they’re opportunities.  Walker’s fascination with rules and their role in the social order has allowed him to find success professionally and personally, from uncovering weaknesses in ne

Microsoft researcher and collaborator combine new mathematical ideas with human-directed AI to resolve a central problem in high-dimensional geometry and probability At a glance Fundamental scientific breakthrough: Researchers resolved the Kannan-Lovász-Si

In the summer of 2010, postdoc Chris White worked with a multidisciplinary team at the Defense Advanced Research Projects Agency, or DARPA, to determine if a new approach to data analysis could prevent violence and protect troops during the war in Afghanistan. At the conclusion o

AI and HPC are transforming data center networking. Modern infrastructures typically consist of a front-end network connecting users, storage, and management systems, and a back-end network interconnecting GPUs and compute nodes. At the largest scale, AI clusters span multiple lo

Convolution is one of the foundational operations in modern machine learning, but it is also a useful example of a broader engineering idea: many apparently specialized tensor operations can be reorganized into well-understood compute patterns that are easier to implement, optimi

By Muhammad Hassan, Mahmoud Mahany, and Mohamed Saleh The semiconductor industry is rapidly transitioning to chiplet-based architectures to meet the performance, bandwidth, and power demands of AI and high-performance computing (HPC). Technologies such as TSMC CoWoS, SoIC, Samsun

City officials are moving beyond isolated pilot projects toward local computing infrastructure and AI. The post Nvidia, Municipal Leaders Frame AI as Critical Urban Infrastructure appeared first on EE Times.

Kiloampere busbars, 2,000 V solar platforms, hollow-shaft robot joints and nanoamp standby budgets are pulling magnetic sensing in directions no single sensing principle can follow. Ask a traction inverter designer, a robotics engineer and someone building a battery-powered water

Quantum investors weigh technical progress against customer demand, commercial timing, and the ability to build lasting value. The post What Makes a Quantum Company Valuable? appeared first on EE Times.

Ising machines solve combinatorial optimization problems by annealing a network of spins. On dense problems, the least expensive hardware step updates all spins at once: stochastic cellular automata (SCA) decide every spin in parallel and read couplings only for the spins that fl

As sequence lengths continue to grow, distributed attention based on sequence parallel (e.g., Ring-Attention) has become a mainstream solution for large language models (LLMs). Nevertheless, inherent communication dependencies limit computation-communication overlap, reducing GPU

The continuous expansion of digital infrastructure, especially data centers, has imposed unprecedented, geographically concentrated stress on existing power grids. However, rigorous causal evidence quantifying how these facilities reshape local power-infrastructure supply remains

We present a toolchain for development of ML models on AWS EC2 F2 cloud FPGA infrastructure from high-level language, implementing a complete Vitis HLS to HDK bridge to AFI workflow that has been absent from the academic literature despite eight years of AWS FPGA availability. We

Rick and Marcia LeCompte loved to watch the Cooper’s hawks that frequented a sweet-gum tree in their yard in Baton Rouge, La. But this October, the hawks were gone. Marcia LeCompte is convinced the birds were scared away by the same flying robots that have been vexing the couple

High motion performance and avoiding structural vibrations rely on smooth trajectory planning. The smoothness requirement depends on the robot's complexity in terms of the number of degrees of freedom and system order due to the flexibility of the joints and links. For industrial

Vision-Language-Action Models (VLAs) enable robots to perform manipulation tasks by leveraging Vision-Language Models (VLMs) to process visual observations and language instructions. However, the action prediction objective itself lacks fine-grained task-specific perceptual super

Projector-based augmented reality enables natural interaction between users and devices by overlaying digital information onto the physical world through projected imagery. A key technical requirement of such systems is accurate positional alignment between the projector and the

Studies of human-robot interaction in dynamic and unstructured environments show that as more advanced robotic capabilities are deployed, the need for cooperative competencies to support collaboration with human problem-holders increases. Designing human-robot systems to meet the

Today, we are expanding our web application firewall (WAF) by adding two new capabilities: full transaction detections and response rules. Full transaction detections evaluate each HTTP request alongside its response to identify evidence of exploitation. It helps answer both "Doe

Update - On October 23, the PHP project published an informational advisory for a related vulnerability, CVE-2026-115560, in PHP's session upload progress feature. It affects fully patched PHP releases, including 8.5.12, and will be fixed in PHP's next regular release.All Cloudfl

On October 19, 2026, at approximately 17:00 UTC, Cloudflare became aware that private keys associated with eight eMudhra intermediate certificates were compromised. These certificates were widely trusted by browsers, operating systems, and also within Cloudflare. Because the keys

Release: datasette-llm 0.1a10 Fixes a bug that affected my agent.datasette.io demo site, where the set of available LLM models was not correctly restricted to just those in the configuration. Tags: datasette, llm

Release: datasette-apps 0.3a0 Datasette Apps adds support for hosting and editing custom HTML applications inside of Datasette. These can be created by hand, but it's easier to create them using LLMs via Datasette Agent. This release improves that integration: Apps can

Release: datasette-agent 0.5a0 This is a significant release of Datasette Agent. The most notable new feature is called the panel - it lets the agent chat open and then sit next to pages from Datasette, for example a table or query page. When combined with datasette-apps

[Updated on 2020-06-17: Add “exploration via disagreement” in the “Forward Dynamics” section. Exploitation versus exploration is a critical topic in Reinforcement Learning. We’d like the RL agent to find the best solution as fast as possible. However, in the meantime, committing

[Updated on 2020-02-03: mentioning PCG in the “Task-Specific Curriculum” section. [Updated on 2020-02-04: Add a new “curriculum through distillation” section.

1.3k points

1.2k points

377 points

369 points

336 points

328 points

300 points

247 points

235 points

240 points

92 upvotes

91 upvotes

69 upvotes

46 upvotes

46 upvotes

19 upvotes

When one goes to Search on lobste.rs there is an Anubis gate first. I understand why it is there for a public rate limiting case, but why is it there for logged-in users? This is an invite only site and if anyone would hammer the server too much they are logged in so they can already be rate limited based on their identity. Am I missing something or is it a configuration problem?

20 upvotes

23 upvotes

⚡ In a hurry? Skip straight to the rules ⚡ Don't Make the Same Mistake I Did I have...

61 reactions

WebMCP is a proposed web standard that, at least to me, looks very promising. I already played around...

32 reactions

The ideal contributor As an open-source maintainer, you imagine the ideal contributor as...

24 reactions

Cabins offer architects opportunities to experiment playfully with architectural form, materials, landscaping, and spatial configurations. Their compact footprints, relatively simple programs, and close proximity to the natural surroundings allow for unconventional design approac

A House in the Green – Nestled between residential buildings and mature trees stood a small former production hall. Originally, it was to be preserved and extended with a senior-friendly apartment, but structural assessments revealed that this would only be possible with extensiv

Some architectural projects only survive as drawings, competition entries, or models in archives. Others were built, but not as first imagined, and the completed building gradually replaced the original proposal in collective memory. Christopher Wren's St. Paul's Cathedral is fam

The Sharjah Architecture Triennial (SAT) has announced the opening programme of its third edition, Architecture Otherwise: Building Civic Infrastructure for Collective Futures. Curated by writer and anthropologist Vyjayanthi Rao, with Associate Curator Tau Tavengwa, the Triennial

House with Rope and Ladder, located in a small village on the southeast coast of Sweden, is a project in two parts: a traditional rendered masonry volume with a gable roof, adhering to the rather strict regulations of the site, and an enclosed circular garden, attached tangential

This week started with World Cities Day, which this year, in line with World Architecture Day, was an occasion to spotlight deficient access to adequate housing, secure tenure, basic services, and affordable living conditions worldwide. Once again, we are faced with a call to app

"We can really bring a hospitality look into a workplace, which is something that our clients have been asking for, for many years," says Mark Harder, a design director at Gensler. The warmth of wood, the texture of leather, and the tactile surfaces found in hotels and restaurant

Located on the Caribbean coast of Central America, Belize is a small country with an intriguing history. It is bordered by Mexico to the north and Guatemala to the west and south, with the Caribbean Sea along its eastern edge. Known as British Honduras for most of its past, it wa

Set on the gentle hillside of Khao Phra Tamnak, the house is conceived through a flexible design diagram rather than a predetermined architectural form. The project begins with the conditions of the place and the lives to be accommodated: a gradual change in land level, the famil

Urban housing in India stands at the intersection of rapid urbanization and a shifting sense of identity. The pressures of land scarcity and market-driven development have transformed housing into a commodity, often prioritizing maximum yield over the fundamental idea of home. In

Tabasco 267 is the renovation of a 90 m² apartment in the Roma Norte neighborhood of Mexico City. CAAM used the existing conditions as the project's raw material, transforming the layout, ceilings, finishes, lighting, and furniture.

International architecture studio SOM has unveiled plans for a new city in Egypt named Alam Al Roum, which will stretch 7.2 kilometres along the Mediterranean coastline. Set to be built near Marsa Matrouh, the 20.6-square-kilometre masterplan features swimmable lagoons alongsi

Canadian architect Brian O'Brian has completed a rural weekend house in Ontario that features a slanted, corrugated-metal facade and an angled roof covered with greenery. The Hemlock House sits on a 72-acre (29-hectare) property in the southern town of Markdale.

designing cities beyond the human   In Landscapes for Adaptation: Evidence from China, architect and Harvard professor Charles Waldheim examines 60 built experiments realized by Chinese landscape architect Kongjian Yu and his practice Turenscape over the past two decades.   Analy

RENNA ENVISIONS FLOATING PLATFORM ON FLORENCE’S RIVER   Designed by Renna Studio and architect Angelo Renna, Firenze Little Island is an architectural proposal for a demountable, seasonal public floating platform on the Arno River in Florence, Italy. Sited along the Pescaia weir,

when utopian visions become reality   Utopias have a tendency to become less interesting as they become real.  At first, they are exhilarating. They overturn conventions, ridicule authorities, and embarrass the present. Their authors anticipate what everyone else has not yet lea

STUDIO IN THE WOODS MARKS 20 YEARS OF HANDS-ON TIMBER LIVE-BUILDS   Marking the 20th anniversary of the architectural live-build summer school Studio in the Woods (SiTW), a new documentary short titled Crossing, Lookout, Horizon records the initiative’s 2026 session hosted at the

A SEVENTEENTH-CENTURY PARISIAN JEWEL BOX   The Prince de Conti hotel reimagines a 17th-century Parisian building, infusing it with contemporary interior design within the historic Saint-Germain-des-Prés district. With just 23 rooms and suites, the four-star hotel has little inter