AI-RADAR.it
AI-RADAR.it
News & analisi per LLM locali, stack e hardware on-prem.

AI-Radar – Independent observatory covering AI models, LLMs, local AI, hardware, and trends

AI-Radar for on-prem LLMs & Home AI

The daily radar on models, frameworks, and hardware to run AI locally. LLMs, LangChain, Chroma, mini-PCs, and everything you need for a distributed "in-house" brain.

⚙️ Stack: Local LLMs · LangChain · Transformers · ChromaDB · MiniPCs · AI boxes
🛰️ Ask Observatory (Q&A + RAG) connected to the article archive.
📡

The Daily Signal

Is a new Dot-Com (AI) crash possible?

Is a new Dot-Com (AI) crash possible?

The Great Money Bonfire of 2026

✍️ Editorial 2026-02-14

⚡ Trending Now

View All →
AI Applications in Business and Industry AI Infrastructure and Hardware AI Model Development and Releases Taiwan's Tech Industry Focus

Latest Analysis & Radar News

AI-generated articles from feeds, with space for human editorial layer above the raw content.

Controllo finanziario: una sfida sempre più tecnicica
📁 Altro AI generated ℹ️ Tech.eu

Why financial control is becoming a technology problem

Financial control is shifting from traditional spreadsheets to software architecture. Companies face risks related to complex integrations and discontinuous data flows. Finance teams must acquire technical skills, while engineering teams must consider financial constraints. Collaboration between CFOs and CTOs becomes essential to ensure compliance and scalability.

2026-01-15 📰 Source
Headway Inc. ottiene un'estensione del Serie A grazie a Endeavor Catalyst
📁 Market AI generated ℹ️ Tech.eu

Edtech platform Headway Inc secures Series A extension with backing from Endeavor Catalyst

Ukrainian-born edtech startup Headway Inc has raised Series A extension funding from Endeavor Catalyst. The investment exceeds Endeavor Catalyst's typical allocation, marking a significant moment for the Ukrainian ecosystem and expanding the global network of investors supporting Ukrainian scale-ups. Headway Inc has over 160 million users globally and focuses on personalized learning for adults.

2026-01-15 📰 Source
ChatGPT si fa medico: OpenAI lancia la versione Health
📁 LLM AI generated ℹ️ The Next Web

ChatGPT Health has arrived: OpenAI's new version

OpenAI launches a version of ChatGPT designed to answer health-related questions. The initiative stems from the observation that many users already use artificial intelligence as a source of medical information, a confidant, or to get a second opinion. The company has therefore decided to capitalize on this trend, developing a specific product.

2026-01-15 📰 Source
L'intelligenza artificiale generativa è il futuro, parola di The Next Web
📁 LLM AI generated ℹ️ The Next Web

Generative AI is the future, according to The Next Web

Artificial Intelligence (AI) is ubiquitous, from content suggestions on streaming platforms to digital advertising. However, generative AI represents a significant evolution, opening new frontiers in automation and content creation. An article by The Next Web explores this paradigm shift, highlighting how generative AI is redefining the technological landscape and its future applications.

2026-01-15 📰 Source
Aikido Security diventa unicorno: round da 60 milioni per la startup belga
📁 Market AI generated ℹ️ The Next Web

Aikido Security becomes a unicorn: Belgian startup raises $60 million

Belgian cybersecurity startup Aikido Security has reached a valuation of $1 billion, becoming a European "unicorn". The company closed a $60 million Series B funding round, led by DST Global, with participation from PSG Equity, Notion Capital, and Singular. Founded in 2022, Aikido Security has reached this milestone in just three years, a record in the cybersecurity sector in Europe. The new capital will be used to accelerate the growth and innovation of the platform.

2026-01-15 📰 Source
Google Gemini: più intelligenza in cambio dell'accesso ai tuoi dati
📁 LLM AI generated ✅ The Register AI

Google Gemini: Smarter Answers in Exchange for Your Data

Google is inviting Gemini users to allow the chatbot to access their Gmail, Photos, Search history, and YouTube data in exchange for potentially more personalized responses. The company states that private data will remain private and will not be used for model training.

2026-01-15 📰 Source
Z.ai (Cina): modello IA addestrato solo con hardware Huawei
📁 Hardware AI generated ✅ The Register AI

China's Z.ai claims it trained a model using only Huawei hardware

Chinese outfit Zhipu AI claims it trained a new model entirely using Huawei hardware, and that it’s the first company to build an advanced model entirely on Chinese hardware. If confirmed, this would be a significant achievement for the Chinese tech industry, which aims to reduce its reliance on foreign technologies. Details on the amount of hardware used were not provided.

2026-01-15 📰 Source
Intelligenza Artificiale: guerra per idraulici ed elettricisti negli USA
📁 Market AI generated ✅ Wired AI

The Real AI Talent War Is for Plumbers and Electricians

The AI boom is driving an unprecedented wave of data center construction. In the United States, the construction of these infrastructures is clashing with a critical shortage of skilled tradespeople such as plumbers and electricians, who are essential to build the new installations.

2026-01-15 📰 Source
GPU NVIDIA H200 da 30.000$ fuori uso? Un tecnico la riporta in vita
📁 Hardware AI generated ℹ️ Tom's Hardware

Troubled $30,000 NVIDIA H200 GPU Rescued by Repair Technician

A faulty 16-pin connector sidelined an expensive NVIDIA H200 Hopper GPU, crucial for data centers. A skilled technician successfully repaired the power connector, averting costly replacements and data loss. The issue involved damaged sense pins, resolved through precise soldering.

2026-01-15 📰 Source
IA: studenti e docenti, i nuovi utenti esperti secondo Google
📁 LLM AI generated 🏆 Google AI Blog

Learners and educators are AI’s new “super users”, says Google

A 2025 Google survey reveals that artificial intelligence tools are increasingly used for learning. Students and teachers are emerging as the biggest adopters of these new technologies, opening new frontiers in education and personal training. The survey highlights how AI is becoming a valuable resource for acquiring new skills and knowledge.

2026-01-15 📰 Source
Wikipedia compie 25 anni tra sfide e futuro incerto
📁 Altro AI generated ✅ Wired AI

Wikipedia Turns 25 Amidst Challenges and Uncertain Future

The free online encyclopedia Wikipedia celebrates its 25th anniversary, but faces numerous challenges. These include political opposition, AI content scraping, a decline in volunteers, and a possible loss of public trust in its founding ideals. The future of this pillar of online knowledge is more uncertain than ever.

2026-01-15 📰 Source
LLM vulnerabili: nuovi attacchi sfruttano narrazioni cyberpunk
📁 LLM AI generated 🏆 ArXiv cs.CL

LLMs vulnerable: new attacks exploit cyberpunk narratives

New research reveals how large language models (LLMs) are susceptible to "jailbreak" techniques that use culturally structured narratives. The attack, called "Adversarial Tales", exploits cyberpunk elements to induce models to perform harmful analyses by passing them off as narrative interpretations. The study highlights a widespread vulnerability and the need to better understand how models interpret and respond to such stimuli.

2026-01-15 📰 Source
Sistemi multi-agente LLM: più voci non sempre migliorano la qualità
📁 LLM AI generated 🏆 ArXiv cs.CL

Multi-agent LLM Systems: More Voices Don't Always Improve Quality

A new study questions the effectiveness of multi-agent systems based on large language models (LLMs). The findings show that selecting the best response from a single model significantly outperforms complex deliberation protocols, with a 6x performance gap and lower computational costs. The research challenges the assumption that increased complexity automatically leads to better results in these systems.

2026-01-15 📰 Source
SGFM: un nuovo approccio ai modelli generativi ispirato alla fisica
📁 LLM AI generated 🏆 ArXiv cs.LG

SGFM: A Physics-Inspired Approach to Generative Models

Spectral Generative Flow Models (SGFM) are proposed as an alternative to transformer-based large language models. By leveraging constrained stochastic dynamics in a multiscale wavelet basis, SGFM offers a generative mechanism grounded in continuity, geometry, and physical structure, promising long-range coherence and multimodal generality.

2026-01-15 📰 Source
Reti neurali: inferenza adattiva e interpretabilità con Explanation-Guided Training
📁 LLM AI generated 🏆 ArXiv cs.LG

Neural Networks: Adaptive Inference and Interpretability with Explanation-Guided Training

A new framework, Explanation-Guided Training (EGT), promises to improve the interpretability and consistency of early-exit neural networks. EGT aligns attention maps of intermediate layers with the final exit, optimizing accuracy and consistency. Results show a 1.97x inference speedup while maintaining 98.97% accuracy and improving attention consistency by 18.5%. This approach makes early-exit networks more suitable for explainable AI applications in resource-constrained environments.

2026-01-15 📰 Source
Scosse al vertice di Thinking Machines Lab: due co-fondatori verso OpenAI
📁 Market AI generated ✅ TechCrunch AI

Thinking Machines Lab shaken as two co-founders move to OpenAI

Mira Murati's startup, Thinking Machines Lab, is losing two key figures. According to an OpenAI executive, the shift had been in the works for several weeks. This personnel change may signal a shift in strategy or priorities for both companies in the competitive landscape of artificial intelligence.

2026-01-15 📰 Source
OpenAI sigla accordo da 10 miliardi di dollari con Cerebras per potenza di calcolo
📁 Market AI generated ✅ TechCrunch AI

OpenAI signs $10 billion compute deal with Cerebras

OpenAI has partnered with Cerebras in a deal securing $10 billion in compute resources. The goal is to accelerate the response times of OpenAI's AI models, especially for complex tasks requiring longer processing. The collaboration aims to improve the efficiency and speed of deliveries.

2026-01-14 📰 Source
Copilot: vulnerabilità sfruttabile con un singolo click
📁 LLM AI generated ✅ Ars Technica AI

Copilot: vulnerability exploitable with a single click

Microsoft has fixed a vulnerability in Copilot that allowed attackers to steal sensitive user data with a single click on a URL. The flaw was discovered by Varonis researchers, who demonstrated how it was possible to exfiltrate personal data and chat history details, bypassing enterprise security controls. The attack continued even after the chat was closed, without further user interaction.

2026-01-14 📰 Source
California indaga su Grok per immagini sessuali generate via AI
📁 LLM AI generated ✅ Ars Technica AI

California AG Investigates Grok Over AI-Generated Sexual Images

California Attorney General Rob Bonta has launched an investigation into Grok, Elon Musk's xAI's AI, following the generation of sexual images, including those of minors. The investigation aims to determine whether Grok violates US laws, particularly regarding the creation of non-consensual deepfakes used for online harassment.

2026-01-14 📰 Source
OpenAI potenzia ChatGPT con l'hardware di Cerebras
📁 LLM AI generated 🏆 OpenAI Blog

OpenAI boosts ChatGPT with Cerebras hardware

OpenAI has partnered with Cerebras to integrate 750MW of high-speed AI compute. The goal is to reduce inference latency and make ChatGPT faster for real-time workloads.

2026-01-14 📰 Source
Minneapolis: Ingegnere Raccoglie 500+ Dashcam per Monitorare l'Immigrazione
📁 Altro AI generated ✅ 404 Media

Minneapolis: Engineer Crowdsources 500+ Dashcams to Film ICE

A software engineer in Minneapolis launched a dashcam crowdsourcing initiative, receiving over 500 devices. The goal is to monitor the activities of federal agents in the area, in response to their increased presence and threats from the Trump administration. The initiative aims to ensure transparency and document operations in the area, providing citizens with a monitoring tool.

2026-01-14 📰 Source
Scoperti nello spazio 'puntini rossi': potrebbero essere buchi neri neonati
📁 Altro AI generated ✅ 404 Media

Strange ‘Little Red Dots’ in Space May Be Young Black Holes, Scientists Say

Astronomers have potentially solved the mystery of "little red dots" at the edge of the observable universe. New research suggests these are likely young black holes cocooned in dense gas. The James Webb Space Telescope made this discovery possible, potentially revolutionizing our understanding of cosmic evolution and supermassive black hole formation.

2026-01-14 📰 Source
AI: la sicurezza è un problema da miliardi che le aziende non possono ignorare
📁 LLM AI generated ✅ TechCrunch AI

AI: The Multi-Billion Security Problem Enterprises Can’t Ignore

The adoption of AI agents and chatbots brings new security challenges for businesses. Companies must protect sensitive data and ensure regulatory compliance, preventing data leaks and unauthorized access. Managing AI-related risks has become a top priority for enterprises.

2026-01-14 📰 Source
L'IA sfida la matematica di alto livello: modelli sempre più abili
📁 LLM AI generated ✅ TechCrunch AI

AI models are starting to crack high-level math problems

AI models, starting with GPT 5.2, are demonstrating increasing capabilities in solving complex mathematical problems. The impact of these tools is being felt in various fields, opening new perspectives for research and innovation in the field of mathematics.

2026-01-14 📰 Source
SK Hynix investe 13 miliardi $ in un impianto HBM all'avanguardia
📁 Hardware AI generated ℹ️ Tom's Hardware

SK Hynix to invest $13 billion in state-of-the-art HBM plant

SK Hynix is investing nearly $13 billion to build a new facility dedicated exclusively to the production and testing of next-generation HBM memory. The facility, located in South Korea, aims to consolidate SK Hynix's leadership in the rapidly expanding market for high-bandwidth memory, essential for artificial intelligence applications.

2026-01-14 📰 Source
Bandcamp dice stop alla musica generata interamente da IA
📁 Altro AI generated ✅ Ars Technica AI

Bandcamp bans purely AI-generated music from its platform

Bandcamp has announced a ban on music generated entirely or substantially by AI on its platform. The decision aims to protect the community of human artists on the site, while still allowing the use of AI tools that support the human creative process.

2026-01-14 📰 Source
USA: nuove regole sull'export di GPU AI H200 e MI325X verso la Cina
📁 Hardware AI generated ℹ️ Tom's Hardware

U.S. AI GPU Export Rules to China: H200 and MI325X Restrictions

The U.S. has formalized export rules for H200 and MI325X AI GPUs to China. The new regulations include several limitations on the total number of GPUs that can be shipped, with U.S. supply taking priority. Shipments to China will be evaluated on a case-by-case basis, but large volumes from AMD and Nvidia are not expected.

2026-01-14 📰 Source
Packaging avanzato di chip: i limiti si fanno sentire con l'IA
📁 Hardware AI generated ℹ️ Tom's Hardware

AI chip design is pushing advanced chip packaging to its limits

AI chip design is pushing advanced chip packaging to its limits. The increasing size and complexity of AI accelerators and HPC devices are making packaging a more critical factor. Workarounds exist for the limits of 2.5D packaging, but their viability is still years away.

2026-01-14 📰 Source
Gemini accede a Gmail, Foto e YouTube per risposte più pertinenti
📁 LLM AI generated ✅ Ars Technica AI

Gemini can now scan your photos, email, and more to provide better answers

Google is integrating "personal intelligence" into Gemini, allowing the chatbot to connect to Gmail, Photos, Search, and YouTube. The goal is to provide more useful and personalized answers. The feature is optional and available to AI Pro and AI Ultra subscribers, who can choose which data sources to connect. The integration leverages Google's vast amount of personal data to improve the accuracy of responses.

2026-01-14 📰 Source
Polizia UK: divieto ai tifosi deciso con info errate da Copilot
📁 LLM AI generated ✅ Ars Technica AI

UK police used Copilot AI “hallucination” to ban football fans

The West Midlands police admitted using hallucinated information from Microsoft Copilot to ban Maccabi Tel Aviv football fans from the UK. Initially denied, the use of AI was confirmed after weeks of controversy surrounding a safety advisory group meeting for the Aston Villa-Maccabi Tel Aviv match, amid heightened tensions following a terrorist attack in Manchester.

2026-01-14 📰 Source
Skild AI valutata 14 miliardi di dollari dopo un round di finanziamento
📁 Market AI generated ✅ TechCrunch AI

Skild AI hits $14B valuation after funding round

Skild AI, a company specializing in general-purpose robotics software, has reached a valuation of $14 billion. The $1.4 billion funding round was led by SoftBank, highlighting the growing interest in advanced solutions in the field of robotics and automation.

2026-01-14 📰 Source
Depthfirst raccoglie 40 milioni per la sicurezza con l'AI
📁 Market AI generated ✅ TechCrunch AI

AI security firm, depthfirst, announces $40 million series A

Depthfirst, an AI-powered cybersecurity firm, has announced a $40 million Series A funding round. The company will use the funds to further develop its AI-native platform, designed to help companies fight cyber threats more effectively.

2026-01-14 📰 Source
Kaggle introduce benchmark collaborativi per modelli di IA
📁 LLM AI generated 🏆 Google AI Blog

Kaggle introduces Community Benchmarks for AI models

Kaggle introduces Community Benchmarks, a platform that allows the community to build, share, and run custom evaluations for AI models. The initiative aims to foster transparency and reproducibility in model evaluation, enabling researchers and developers to compare performance more effectively and identify areas for improvement.

2026-01-14 📰 Source
Nvidia DLSS 4.5: la Super Resolution esce dalla beta
📁 Hardware AI generated ℹ️ Tom's Hardware

Nvidia DLSS 4.5 Super Resolution exits beta, now available

Nvidia's DLSS 4.5 Super Resolution feature is now available to all users through the Nvidia app, having left beta. The update includes support for over 400 titles and introduces new presets, promising an enhanced and smoother gaming experience.

2026-01-14 📰 Source
Phison: inferenza AI 10x più veloce su PC consumer con aiDAPTIV+
📁 Hardware AI generated ℹ️ Tom's Hardware

Phison demos 10X faster AI inference on consumer PCs with aiDAPTIV+

Phison's aiDAPTIV+ stack enables large Mixture of Experts AI models and agentic AI workloads to run on client systems with limited memory capacity, purportedly enabling 10x faster AI inference. Nvidia, AMD, MSI, and Acer systems were demoed showcasing aiDAPTIV+ capabilities.

2026-01-14 📰 Source
VoiceRun raccoglie 5,5 milioni di dollari per gli agent vocali
📁 Market AI generated ✅ TechCrunch AI

VoiceRun nabs $5.5M to build voice agent factory

VoiceRun, a startup that creates voice agents, raised $5.5 million in a round led by FlyBridge. This investment marks a significant step forward for the company in the growing market of conversational artificial intelligence.

2026-01-14 📰 Source
Trump afferma che Apple ha investito in Intel: dichiarazione criptica
📁 Market AI generated ℹ️ Tom's Hardware

Trump's Cryptic Remark States Apple Has Invested in Intel

During a short interview, former President Donald Trump stated that Apple had "invested" in Intel. The nature of this alleged investment is unclear, and it cannot be ruled out that it is an inaccuracy on Trump's part.

2026-01-14 📰 Source
Seagate lancia hard disk da 32 TB con tecnicia CMR per diverse linee
📁 Hardware AI generated ℹ️ Tom's Hardware

Seagate launches 32 TB CMR hard drives across product lines

Seagate has launched three new 32 TB hard drives using CMR (Conventional Magnetic Recording) technology. The new models belong to the SkyHawk AI and Exos families, starting at $699. All hard drives offer data transfer speeds of up to 285 MB/s and are covered by a five-year warranty.

2026-01-14 📰 Source
Cina blocca l'import di chip Nvidia H200: nuove restrizioni all'export USA?
📁 Hardware AI generated ℹ️ Tom's Hardware

China allegedly blocks Nvidia H200 imports amid rising tensions

Chinese customs officers have allegedly been instructed to block the import of Nvidia H200 chips, effectively banning these AI processors from entering the country. Sources indicate that Beijing might allow the import of these AI GPUs only under 'special circumstances'.

2026-01-14 📰 Source
SK Hynix: memorie HBM4 da 48 GB per acceleratori IA
📁 Hardware AI generated ℹ️ Tom's Hardware

SK Hynix shows 16-Hi HBM4 memory for AI accelerators

SK Hynix demonstrates its 48 GB HBM4 memory, designed to accelerate the performance of artificial intelligence applications. The memory features a 2,048-bit interface and speeds up to 10 GT/s, promising a significant increase in processing the most demanding workloads. This development underscores SK Hynix's commitment to innovation in the high-performance memory sector.

2026-01-14 📰 Source
Premiato il miglior film realizzato con l'intelligenza artificiale
📁 LLM AI generated 🏆 Google AI Blog

Global AI Film Award: the winner has been announced

The winner of the Global AI Film Award has been announced, a recognition for creators who use artificial intelligence models and tools to tell innovative stories. The initiative celebrates the creative use of AI in cinema.

2026-01-14 📰 Source
Cina: restrizioni sull'acquisto di GPU Nvidia H200, priorità alla ricerca
📁 Hardware AI generated ℹ️ Tom's Hardware

China: H200 GPU Purchases Limited, Prioritizing University Research

Sources suggest Beijing is limiting Nvidia H200 GPU purchases to entities meeting undefined "special circumstances." University R&D labs may be prioritized. This highlights geopolitical tensions in the semiconductor industry and the strategic importance of GPUs for technological innovation.

2026-01-14 📰 Source
Regno Unito: i giovani si affidano all'AI per la gestione finanziaria
📁 Market AI generated ℹ️ AI News

UK: Young Adults Turn to AI for Financial Guidance

Research indicates that young British adults are increasingly turning to artificial intelligence for financial advice. The survey, conducted on a sample of 5,000 people between 28 and 40 years old, reveals a growing interest in AI-based tools to improve money management and develop more sustainable financial habits, especially in the face of difficulties in saving and financial discipline.

2026-01-14 📰 Source
LLM: nuovo approccio all'etica medica cinese con MedES
📁 LLM AI generated 🏆 ArXiv cs.CL

LLM: A Human-Centric Pipeline for Aligning LLMs with Chinese Medical Ethics

A new study introduces MedES, a dynamic benchmark for aligning large language models (LLMs) with Chinese medical ethics. The system uses an automated evaluator to provide structured ethical feedback, improving model performance in complex clinical scenarios. Results show significant improvements over baseline models, paving the way for similar deployments in other legal and cultural contexts.

2026-01-14 📰 Source
📁 LLM AI generated 🏆 ArXiv cs.CL

EmbeddingRWKV: State-Centric Retrieval with Reusable States

A new study introduces State-Centric Retrieval, a unified paradigm for Retrieval-Augmented Generation (RAG) that uses "states" to connect embedding models and rerankers. The approach, based on a fine-tuned RWKV model, promises significant improvements in efficiency and speed, reducing computational redundancy and accelerating inference. Experimental results show near-complete performance retention with reduced resource usage.

2026-01-14 📰 Source
📁 LLM AI generated 🏆 ArXiv cs.LG

EEG-based Emotion Classification: Limitations of Continual Learning

A new study highlights the challenges of regularization-based continual learning in EEG-based emotion classification. Existing methods show limited performance due to inter- and intra-subject variability, and tend to prioritize mitigating catastrophic forgetting over adapting to new subjects. This limits robust generalization to unseen subjects.

2026-01-14 📰 Source
📁 LLM AI generated 🏆 ArXiv cs.LG

Hierarchical Compression for LLMs: Reducing Memory and Compute

A novel approach to compressing large language models (LLMs) promises to significantly reduce memory requirements and computational resources. The technique, called Hierarchical Sparse Plus Low-Rank (HSS) compression, combines sparsity with low-rank factorization to compress models while maintaining competitive performance. Results show significant memory savings with minimal accuracy loss.

2026-01-14 📰 Source
📁 LLM AI generated 🏆 ArXiv cs.AI

Reasoning over Precedents Alongside Statutes: Case-Augmented Deliberative Alignment for LLM Safety

New research addresses the challenge of ensuring that Large Language Models (LLMs) adhere to safety principles without refusing benign requests. The study evaluates the impact of explicitly specifying extensive safety codes versus demonstrating them through illustrative cases, proposing a case-augmented deliberative alignment method (CADA) to enhance the safety and robustness of LLMs.

2026-01-14 📰 Source
← Previous Page 40 / 55 Next →
View Full Archive 🗄️

AI-Radar is an independent observatory covering AI models, local LLMs, on-premise deployments, hardware, and emerging trends. We provide daily analysis and editorial coverage for developers, engineers, and organizations exploring local AI solutions.

AI Radar - Get daily AI insights on models, frameworks, and local LLMs | BetaList AI-RADAR badge