All published intelligence, sovereign stack research, ecosystem news, and AI foundation model reviews.
OpenAI завезли GPT-6.1 Sol Ultrafast. Команда постепенно запускает его в API, Codex и ChatGPT Work. Скорость — до 8 раз выше, чем у обычного Sol. Но, он дороже обычной Astra. Хотя за такую скорость цена всё ещё кажется оправданной. Через API им может пользоваться кто угодно, а в…
Sakana предложила новую архитектуру памяти: нейросеть сама решает, когда вспоминать, а когда думать Sakana AI выпустила работу о новой рекуррентной архитектуре Continuous Memory Machine (CMM). Sakana предлагает разделение памятей по ролям и то, как они взаимодействуют на: -…
OpenAI fired three safety researchers who helped investigate the Hugging Face hack. In an open letter, they warn that the firings are scaring remaining staff and eroding safety culture. OpenAI claims they violated policies but won't say how. The article OpenAI's safety crisis…
Anthropic официально запрещает "оскорбительное или жестокое обращение" с Claude Дарио Амодеи объясняет позицию так: «Мы не знаем, обладает ли claude сознанием». поэтому мы приняли определённые меры, чтобы у моделей был "хороший опыт"». >claude имеет право закончить разговор,…
📌 Кейноут Windows and Surface event На осенней презентации 7 октября Microsoft рассказала о своих новинках и планах. Встроенная в Windows 11 песочница Microsoft Execution Containers (MXC), которая на уровне системы ограничивает, к каким файлам и сетям допущен агент, вышла из…
OpenAI is standing firm on its decision to fire three safety researchers after an investigation found they committed "a significant breach of trust." In a post on X on Friday, the company said Jasmine Wang, Tomek Korbak and Mikita Balesni were dismissed for violating "clear…
Astrophysicist Brice Ménard of Johns Hopkins University used Anthropic's Claude Science to map the entire sky in ultraviolet light for the first time. AI agents downloaded data from multiple space missions, calibrated it, and filled in gaps using inpainting. Predictions averaged…
Ever since people first seriously contemplated giving machines an intelligence modeled on our own, there has never been any question that they would, like us, be able to say no. The sci-fi canon is full of stories of robotic disobedience. Most of these capers are, of course,…
OpenAI exposed a Russian and an Iranian influence operation and banned the accounts involved. The Russian "Dark Clark" campaign spread disinformation across Latin America and drew reactions from politicians, earning the first category 5 out of 6 rating in OpenAI's reporting…
I'm John Peterson. I'm chief technology officer at Sophos. And in my role, I'm responsible for all of the R&D for the company. So, all the [music] products that we build, all the threat intelligence that goes into those products, and the support organization that supports those…
За пару дней Anthropic выпустили много нового и начали выдавать $ подписчикам уровня Max/Team на API каждый месяц Итак, подписчикам Max и Team теперь выдают ежемесячные кредиты на API: $100, $200 и до $500 на команду. Claude теперь может создать из вашей переписки ролики и…
🤖 ИИ выдал 284 ответа, которые противоречили подтверждённым фактам Digital-агентство Verndale проверило, как шесть популярных ИИ-систем рассказывают о программах одного американского университета: более 7200 ответов на 100 вопросов в ChatGPT, Gemini, Google AI Overviews, Google…
Discover how Sophos uses OpenAI’s Daybreak to cut cyber-threat investigation time by 96% and automate 52% of MDR cases while preserving human oversight.
✔️ Meta*, DeepMind и Isomorphic Labs присоединились к проекту симуляции живых клеток Некоммерческая научно-исследовательская организация Biohub семейного фонда Цукербергов расширила программу Virtual Biology Initiative для симуляции клеточных процессов in silico. К проекту…
💸 Anthropic начала раздавать API-кредиты подписчикам Claude Max Теперь в зависимости от уровня подписки пользователям дают отдельный ежемесячный бюджет на API-вызовы Claude — можно тратить его из собственных приложений, скриптов, CI-сборок и других проектов. При этом лимиты…
The math apocalypse has begun. Scott Aaronson calls this the biggest day in mathematical history. A global coalition of mathematicians formed together to oppose AI math. Only humans allowed. No clankers. Now, Anthropic worried that some of these strongly worded letters might…
Human communities are governed by normative systems: shared standards that produce \textit{norms} dictating acceptable behavior, enforced through community sanctioning. Aligning increasingly autonomous AI systems with these norms is a central alignment challenge, complicated by…
Behavior cloning (BC) in non-Markovian environments is a challenging problem because policies have to reason over contextual information over long horizons. Existing policy architectures rely on recurrent or attention-based mechanisms to capture long-term dependencies. However,…
Aerial vision-and-language navigation (VLN) agents are typically trained on detail-rich, trajectory-aligned commands, whereas users issue short, intent-driven instructions; on a frozen OpenFly navigator, this \emph{instruction gap} drops success rate (SR) from $31.03\%$ to…
Agentic AI systems can improve by searching longer, receiving additional support, or modifying how they propose and verify outputs. A performance score does not distinguish these mechanisms. We compare these changes through bounded verification with hidden terminal randomness. A…
Tool-using agents repeatedly carry observations whose useful content can be much smaller than their original payload. We study agent-controlled forgetting: the acting model selects previously observed tool results, replaces each with a short note at its original position, and…
Powerful misaligned AI models might recognize alignment evaluations and strategically behave well on them, rendering direct audits uninformative. However, distilling such a model into a weaker benign student places the teacher in a Distillation Double Bind: if misalignment…
As AI systems increasingly interact with people and make decisions about them, understanding human interpretations becomes an important part of developing human-centered AI. Conventional machine learning and AI systems are largely developed under the assumption that a single…
Game commentary is an open-ended generation task requiring multimodal perception, strategic reasoning, and contextual knowledge. Existing AI-Generated Game Commentary (AI-GGC) studies remain fragmented across games, modalities, and evaluation protocols, while overlap-based or…
A verified reference solution provides a correct trajectory for training a reasoning model. Alternatively, a prefix of the reference can guide the model in generating a trajectory of its own. How much reference guidance should we provide? We study this question through prefix…
Reinforcement learning environments are now a primary lever for improving large language model (LLM) capabilities in post-training, yet most agentic benchmarks remain static: the world moves only when the agent acts, the reward is a terminal verdict, and the pass bar is set…
Most human-agent interaction today remains text-based. Natural language can impose cognitive overload, ambiguity, information chaos, and slow input for complex tasks; ephemeral generative UIs can present structured information and guide users toward task completion. We propose…
Computer-use agents are capable of completing complex tasks, increasing the use of automatic judges to determine success, either for training or for evaluation without human involvement. Despite their flexibility, their reliability on long tasks spanning multiple applications…
Safety-aligned language models often refuse a harmful request stated directly but answer the same request inside a role-play or narrative wrapper. We measure this vulnerability across languages and registers: attack success on Qwen3-1.7B is already 89.4% in English and 93.0% in…
Heuristic search for a plan can store exponentially many states, even when its heuristic is almost perfect. We instead learn search control, one specification per domain, written as an indexical policy: a generalized policy with registers that hold objects and modes that…