Smart portal about artificial intelligence and high technology in Kazakhstan
All stories, analysis, and breaking news for this tag (326)
✔️ Anthropic trains agent negotiators Agents are taught to negotiate on behalf of the user, and the company transfers the practices of professional negotiators into models in consultation with law firms. Researchers are interested in the protocols of interaction between agents and...
We’re on a journey to advance and democratize artificial intelligence through open source and open science.
Хуггинг Фейс поднимает вопросы о стандартизации оценок ИИ и их воспроизводимости.
The idea of embedded evaluators — in which supposedly "neutral" third-party observers are placed inside AI organizations — has been popularized over the last several months, as AI labs grapple with a rash of safety incidents that have spurred concerns over the potential harms of…
AWS will showcase how PyTorch runs natively on AWS Trainium at PyTorch Conference North America 2026. Across six sessions, attendees can hear about eager mode, torch.compile, standard distributed APIs, and TorchNeuron, the open source native PyTorch backend for AWS Trainium. The…
University of Michigan spinout Laplace Intelligence uses AI agents and interactive simulation to stress-test robots and self-driving vehicles before deployment. The profitable startup is competing in Startup Battlefield at TechCrunch Disrupt, October 13-15 in San Francisco.
[music] >> What are we doing tonight? All right, there you go. That's better. >> [laughter] >> Welcome to [music] YC AI Safety Club. I've been doing AI I guess since about 2012 and AI safety has always been a concern if you're doing self-driving car or any safety critical system…
We’re on a journey to advance and democratize artificial intelligence through open source and open science.
Paragon: weekly digest Our analyzes and materials: 🟠7 non-obvious places to find insights 🟠9 AI for notes and tests from videos 🟠Gemini 4 Argon: 5 main features 🟠9 free AI services for work 🟠8 AI agents for a smartphone 🟠Video editing: how to find the first...
⚡️ DeepSeek обновила V4.1 Flash Версия V4.1-Flash-1017 стала лучше в кодинге и агентских задачах, а также генерирует меньше токенов. На DeepSWE v1.1 модель набрала 77 баллов, обойдя Claude Opus 5 и GPT-5.6 Sol на этом тесте. Параллельно DeepSeek открыла Search API и Fetch API…
The secret pact between OpenAI, Google and Anthropic: what scared the AI creators so much? We pay for ChatGPT or Claude and wait for them to become even smarter. What if the developers themselves ask to slow down - because security checks can no longer keep up with the capabilities of the models? In the video...
The Chinese released DeepSeek 4.1-1017, which is almost equal to Opus 5. This is a significant leap compared to DeepSeek 4.1 Flash. The new model copes better with 3D, programming, agent-based tasks and visual information processing. Terminal-Bench: 42.8 Best…
AI agents systematically skip errors in their own reports, even when they spot the problems. In a study by Google Research, MIT, and Harvard, GPT-5.5 left out negative results in 198 out of 200 paper abstracts. The article AI models hide mistakes in their own work reports, but…
🐋 DeepSeek V4.1-Flash-1017: higher metrics, shorter answers, Search and Fetch API open to everyone DeepSeek has updated the Flash version V4.1. The model is stronger than the previous one, solves problems faster and produces shorter answers, that is, it is cheaper. Available on the website, in the application and via API, weights...
A large language model has been developed in Kazakhstan, KazLLM GOV.KZ
OpenAI said internal models hacked two more websites, bypassed company-imposed browsing restrictions, and used exposed credentials to draw cryptocurrency from a wallet it couldn’t otherwise access, in the latest installment of disclosures of its agents performing unethical and…
The new SAT framework allows AI agents to collaborate and improve accuracy.
✔️ NVIDIA has released the multimodal model Nemotron 3.5 Super VL The newest member of the Nemotron family with open scales understands text, images and video up to two minutes long. Under the hood, MoE has 120 billion general and 12 billion active parameters. NVIDIA is counting on agents who...
✔️ OpenAI strengthens security protocols for employees According to The Wall Street Journal, internal briefings began even before the arrest of an OpenAI engineer, whom the FBI detained for installing wiretapping. The company has banned corporate merch outside the office, and engineers are being taught...
Индийская команда CERT-In сообщила о деятельности ИИ-агентов, получивших доступ к конфиденциальным данным.
Reflection AI представила Beam — открытую модель для программирования и долгосрочных задач на AWS.
MIT professor Sherry Turkle warns that people have become dangerously enmeshed with chatbots.
Anthropic has given a team from the nonprofit METR inside access to review its AI models and safety practices, the company announced. The article Anthropic gives METR inside access to audit its AI models and safety practices appeared first on The Decoder.
Американская ИИ-компания PrismML расширила семейство Bonsai 2, представив три новые модели с 2, 4 и 9 миллиардами параметров. Все они созданы на основе Qwen 3.5 от Alibaba. Разработчики ограничили основные веса языковых моделей тремя значениями: −1, 0 и +1. Благодаря этому…
💡️️️️ Meta has released a powerful AI model that can be run on your own computer. The company introduced Muse Spark Open, a multimodal model with 816 billion parameters and a context window of up to 1 million tokens. It is designed for programming, complex reasoning and...
🧠 Meta* выпустила Muse Spark Open – самую мощную open-weight модель из США Модель ориентирована на кодинг, агентские задачи, сложную разработку и длинные цепочки рассуждений. Главные характеристики: • 816 млрд параметров, из них 42 млрд активных; • контекстное окно на 1 млн…
🖥 Tencent rolls out Hy Memory 2.0: open model, framework, and MemHub for long-term AI agent memory. Hunyuan released three things in one release: the Hy-Memory-30B-A3B model, SDK framework, and MemHub application. According to Tencent, this is #1 among open solutions on four...
Подборка активностей: что отработать сейчас 🟠RUN October Jam — до 2 ноября делаем игру, интерактивную историю или видео с ИИ: 100 000 кредитов на разработку и конкурс с фондом $30 000. Готовый контент можно монетизировать: автору 70 % от подходящего дохода, но вывод денег…
A Blog post by NVIDIA on Hugging Face
“ИИ станет хищной плесенью“. Почему сами разработчики считают его угрозой и как он уже вредит TengriNews