Все опубликованные аналитические материалы, исследования суверенного стека, новости экосистемы и обзоры моделей ИИ.
In the UK, public healthcare staff report turning to machine translation (MT) - predominantly Google Translate (GT) - to communicate with patients across language barriers. Though intended to support their duty of care, potentially uninformed reliance on MT in such contexts…
The operation of dynamically polarized targets in nuclear physics experiments relies on continuous tuning of the microwave frequency to compensate for radiation damage and evolving material properties, a task that is traditionally performed through manual trial-and-error by…
Text-to-image generation enables users to explore several images generated from the same prompt. For these generated images to be useful, each one must reflect the user's preferences, measured by a learned reward, and differ visually from the others to maintain diversity.…
Large language models spend the same amount of computation on every token they generate, regardless of how difficult each token is to produce. Methods such as speculative decoding and model routing are built on the premise that much of this computation is unnecessary, yet the…
Agent harnesses make many small, typed decisions per task: which model to call, which tool to use, whether retrieved text is relevant, whether an input carries an injection. System-1 decision models answer such questions in a single forward pass with class probabilities,…
Using a frontier model like Claude Opus as a meta-agent to generate terminal tasks and verifiers for RL training is increasingly common. Yet a runnable Docker image and executable test suite do not guarantee a faithful end-to-end pipeline for terminal agent training. We present…
Streaming GPU solvers for entropic optimal transport (EOT), such as FlashSinkhorn, avoid storing the dense kernel but still evaluate all $n\times m$ point pairs in every Sinkhorn iteration. We present \textbf{FlashSinkhorn~2} (FS2), a solver for squared-Euclidean cost on…
Training-free safeguards for text-to-image generation often rely on a reusable safety signal, such as an unsafe direction or global toxic subspace, applied broadly across prompts. We provide a controlled geometric analysis of this global-unsafety assumption and reveal a…
Societal resilience research relies on access to useful and actionable data, which motivates our main research question: Can annual reports, processed at scale with LLMs, provide a useful signal about how companies disclose their response to AI? We test this by applying a…
LLM judges are increasingly used to assess whether AI outputs meet workplace requirements, but agreement on response rankings does not establish agreement on acceptance rates or occupational aggregates. We introduce O*NET-BENCH, an audit suite derived from an existing survey of…
World models (WMs) simulate the transition dynamics of environments, enabling agents to plan over the consequences of their actions. In text-based environments, fine-tuning a Language Model (LM) to serve as a WM has emerged as a dominant paradigm. However, despite the widespread…
Recent years have seen the employment of a plethora of machine learning (ML) models in high-stakes domains, but they remain largely opaque to the practitioners who act on their predictions. While post-hoc explanation methods offer a lens into this model behavior, wielding them…
Campus AI tutors based on retrieval-augmented generation (RAG) must ground answers in assigned course materials while keeping textbooks and student dialogue on institutional infrastructure. We present CourseChat, an on-premises, multi-course RAG tutor for undergraduate business…
As powerful AI systems reach and sometimes surpass the abilities of human experts across a range of cognitively demanding tasks, the problem of accurate oversight and supervision of these systems has become increasingly urgent. One promising approach is AI debate, which seeks to…
Developing cognitive models of algorithmic reasoning from behavioral data is a central problem in cognitive science that challenges current methods. Traditional approaches to cognitive modeling are interpretable and benefit from human expertise, but lack flexibility and…
Natural Visibility Graph (NVG) based analysis characterizes network traffic through topological descriptors reflecting different structural properties. However, not all descriptors contribute equally to cyber-attack classification, and extracting a large metric set can increase…
Long-horizon research agents must decide both how to investigate and what to investigate next as evidence accumulates. This is hard to learn because such decisions are sparse in long execution traces, and their consequences may emerge several investigations later. We introduce…
World action models (WAMs) have emerged as a promising paradigm for robotic control by jointly predicting future visual dynamics and actions from an initial observation and instruction. However, existing WAMs struggle with long-horizon prediction, as generating dense video…
Interactive world models are increasingly capable of generating environments and acting within them, yet deliberately editing an existing executable world remains underexplored. We formulate world editing as intervening on an existing world while preserving properties that…
Обсудили, почему в ИТ-сфере китайские производители на голову выше европейских брендов и как обновление софта может улучшить машину даже через год после покупки.
⚠️ «Культура OpenAI сломана» — один из старейших сотрудников компании ушёл из-за опасений за безопасность ИИ Дэвид Робинсон проработал в OpenAI три с половиной года и руководил подготовкой отчётов о безопасности для крупных запусков. Теперь он увольняется, потому что считает…
StarSkirmish ставит ботов для StarCraft, созданных ИИ, друг против друга, а также против ботов, созданных людьми. GPT-6 Astra от OpenAI и Claude Opus 5.5 по сути оказались внизу как лучшие боты, созданные ИИ, но они не смогли обогнать Stardust, лучшего бота, созданного человеком. В пятницу,…
All right, so we have some sobering news today from the world of AI. A key figure at OpenAI quits, saying that the culture at OpenAI is broken. Specifically talking about the safety approach they're taking with building AI. Another employee that still works at OpenAI, who's a…
DevDay has passed: OpenAI, new Codex, updated Workspace, and of course, 24/7 bots and an agent with its own computer . Inside the OpenAI ecosystem, Anthropic releases the little brother, Claude 3.5. 30%faster, cheaper, and almost as smart. An unexpected frontier model release…
Бывший сотрудник OpenAI Дэвид Робинсон, проработавший в компании 3,5 года, заявил, что разработчики искусственного интеллекта двигаются слишком быстро и не уделяют безопасности столько внимания, сколько необходимо. Он участвовал в создании системы оценки рисков OpenAI и…
💸 OpenAI возвращает ChatGPT Pro за $200 в месяц — но лимиты будут в два раза меньше Подписка снова обойдётся в $200 (около ₽16 900) в месяц, однако пользоваться моделями можно будет заметно меньше: OpenAI решила сократить прежние лимиты вдвое. В компании объясняют это ростом…
Будут ли в Алматы сильные землетрясения в ближайшие три года? ИИ проанализировал историю толчков и объяснил, что известно о рисках.
Пост в блоге Microsoft на Hugging Face
✔️ Anthropic открыла моды для Claude Code Моды – небольшие функции на TypeScript, меняющие работу ассистента в CLI и десктопном приложении. Там, где хуки лишь получали управление по событию, мод перехватывает его и переписывает: подменяет промпт до отправки модели, блокирует или…
☢️ Нейронкам устроили проверку на человечность — и ответы сильно разошлись Исследователи протестировали 10 ИИ-моделей на моральной дилемме: допустимо ли применить насилие к одному человеку, если это необходимо, чтобы предотвратить ядерный апокалипсис. Единственной моделью,…