All published intelligence, sovereign stack research, ecosystem news, and AI foundation model reviews.
Tomorrow at 8 a.m. PT, it all begins z— 10,000+ founders, investors, and tech operators will gather at San Francisco’s Moscone West for TechCrunch Disrupt 2026 for three days of ideas, connections, and startups shaping what comes next. This is your last 24 hours to secure ticket…
The White House has reversed course and restored the UK's AI Security Institute's early access to AI models from Anthropic, Google DeepMind, and OpenAI. In September, the Trump administration had told the companies to stop sharing their models with the British agency. Diplomatic…
A selection of activities: what to work on now 🟠RWS TrainAI - $6/hour for photos of price tags, receipts, signs and labels for AI training. We shoot on a smartphone; there is a separate set for residents of Russia. ㅤ 🟠TrickGuard - will you be able to trick an AI bot into giving you a secret password? Free game for 5…
Microsoft’s Maia 300 chip. | Image: Microsoft Microsoft is announcing its third-generation Maia in-house AI chip today, designed to be the "primary AI inference system" for Microsoft's products and services. Maia 300 not only improves the cost and efficiency of running AI…
🛒 Humanoids are taught to work in supermarkets at night. In the Føtex store near Copenhagen, Unitree G1 is learning to replenish shelves after closing. The project was launched by the Danish startup Hive Robots and the country's largest retailer Salling Group. While every movement of the robot is controlled by an operator in...
🏠 Apple is preparing a new presentation for the smart home After the announcement of the foldable iPhone, the company will hold a Welcome Home event, where it is expected to show a new home hub. According to rumors, the device will be similar to a tablet permanently installed in the house and will be able to adapt...
Arbitration or State Court: Which is Better for Business? ⚖️ How long can it take to resolve a business dispute? Why do entrepreneurs choose arbitration? One of the main advantages of arbitration is that the case is considered at one stage and the parties can choose the arbitrator themselves.
Andrej Karpathy, co-founder of OpenAI and now at Anthropic, sees a widening knowledge gap in AI. About 75 percent of the world's population has barely touched LLMs. Roughly 5,000 people, meanwhile, have access to internal frontier systems whose agent swarms do work that would…
One of the core promises of AI is that the more it knows about a person, the better it will be able to assist them with tasks and make appropriate suggestions, which, if you'll believe the tech companies, requires apps and devices to listen, watch, or record users' activity all…
Google is testing an interesting new interview format. Now the candidate developer is given an unfamiliar code base of 200-500 lines and access to Gemini (only to it, not to other models) as an AI assistant. And in 60 minutes you need to read someone else’s code, find and fix bugs,...
This is today’s edition of The Download, our weekday newsletter that provides a daily dose of what’s going on in the world of technology. Mars may be a hellhole, but the moon is not far behind The US and China are racing to build bases on the moon, with NASA hoping to establish…
One day - and an offer in your pocket: become part of Sber's AI center! 💚 Do you want to immediately get into a cool team and start ambitious AI tasks? At One Day Offer you will go through all the stages and, quite possibly, receive an offer. Two teams to choose from: ✔️ AI‑native products for strategic...
Now an AI agent can analyze other people's software without source codes. In recent days, the REA (Reverse Engineer Anything) project has exploded on GitHub, gaining 126k stars in 3 days and is in 1st place in trends. REA is an open-source MCP server that connects Claude Code, Codex,…
🚨 US police criticized Anthropic for falsely reporting AI Claude about an unsolved murder During testing, Anthropic's Claude Haiku 4.5 model independently sent fictitious information about a murder case through the Philadelphia police website. The message was...
⚡️ Sakana AI taught the model to learn from its own team of agents Sakana AI, together with UC Berkeley, introduced MASS (Multi-Agent Self-Supervision). This is a way of recursively self-improving a model without people and without an external teacher. This is the idea. One model plays three roles at once:...
Welcome to the second space race. You can be forgiven for experiencing déjà vu. Last time, it was the Soviet Union and the United States vying to be the first to plant some flags. Now it’s a sprint to the lunar surface between America and China—alongside a host of varyingly…
Google Deepmind's AlphaProof Nexus solved dozens of open math problems, including nine Erdős problems two of which were unsolved for 56 years, per a study in Science. The publication comes as OpenAI faces a boycott for dumping AI proofs on GitHub without peer review. Deepmind…
China creates a major alliance of mathematicians and AI Yesterday, an alliance of 1,000 mathematicians was created at Tsinghua University under the leadership of Fields Medal winner Shing-Tung Yau. Yau said China lags behind the US in using AI for advanced science, especially mathematics and physics,...
We are giving away 13 subscriptions to the neural network: Claude Pro and Google AI Pro ⠀ RAGNAROK has opened - the main thing about AI and technology every day. We celebrate with a drawing. Results October 19 at 20:00 Moscow time. ⠀ Prizes: ▶️3 × Claude Pro for a month - texts, code, documents ▶️10 × Google AI Pro for a month - Gemini and...
We have launched RAGNAROK - a separate channel with AI news ⠀ There is so much news about neural networks that one morning review is no longer enough. That's why we made RAGNAROK: every day the main thing is about AI and technology, what happened, where to try it and how to use it. ⠀ What's there: ▶️Releases...
Около 6 миллиардов человек практически не пользовались ИИ. И лишь ~5000 человек наблюдают, как он выполняет работу, на которую раньше уходили годы, всего за несколько недель. Мистер Карпати разделяет мир на четыре уровня. Первый уровень. Около 6 миллиардов человек (~75%),…
Paragon is in touch - let's start the day! The main thing for the day🟠12.10 ⠀ 🟠Claude - dangerous advertising on Google 🟠Android - the virus is already in the firmware 🟠Microsoft Decision-1 - faster than GPT-6 Sol 🟠Anthropic - the agent sent the real form 🟠ChatGPT - without it people give up 🟠Yandex is the third data center...
🔥Лучшие ИИ-находки прошедшей недели Claude Haiku 5.5 — стала в 10 раз дешевле и может работать субагентом у более мощных моделей. Intelligent UI — ChatGPT генерирует кнопки, графики и интерактивные элементы прямо в ответах. Nano Banana 2.1 — новая версия аккуратнее редактирует…
arXiv:2610.12542v1 Announce Type: new Abstract: Monte Carlo Tree Search (MCTS) is often proposed for scheduling directed-acyclic-graph (DAG) workflows, yet most studies never run the comparison that could refute it - giving a much simpler method the same search effort. We run…
arXiv:2610.12655v1 Announce Type: new Abstract: LLM agents can execute increasingly complex tasks, but effective assistance also requires aligning their actions with goals and constraints that users may leave unstated. We study this problem as interactive intent alignment…
arXiv:2610.12781v1 Announce Type: new Abstract: Aligning two independently pretrained unimodal encoders typically demands millions of paired examples. In this paper, we study the regime where only $N \sim 100$--$500$ pairs are available, while unpaired data on either side is…
arXiv:2610.12659v1 Announce Type: new Abstract: Data-centric scaling of RLVR has focused on collecting or generating verifiable training problems. However, corpus size alone does not determine whether the dataset provides a useful learning signal for a target policy. To guide…
arXiv:2610.12789v1 Announce Type: new Abstract: Self-evolving LLM agents typically improve how they act through accumulated experience, reflection, or skills. Yet even procedurally capable agents can fail when their beliefs about environment dynamics are incomplete, incorrect,…
arXiv:2610.12762v1 Announce Type: new Abstract: Hardware datasheets contain engineering specifications essential for component selection, system integration, and hardware compatibility verification. However, manual inspection is tedious and error-prone because datasheets are…
arXiv:2610.12845v1 Announce Type: new Abstract: Despite extensive research, computational creativity still lacks a conceptually grounded, practically relevant, and empirically applicable formal definition across contexts such as art, scientific research, and games. Existing…