All published intelligence, sovereign stack research, ecosystem news, and AI foundation model reviews.
Memory poisoning attacks on LLM agents typically assume an external adversary who plants content in the agent's persistent memory to steer its behavior. We instead study, with no adversary involved, whether a misaligned agent can write a goal it cannot yet act on to persistent…
Financial LLM agents are often evaluated by comparing their end-to-end returns with those of a baseline and testing the paired difference against zero. This measures whether deploying the agent changes realized performance, but it does not isolate event-selection skill. An agent…
Synthesizing realistic graphs at scale is vital when the graphs of interest are large and real-world samples are limited or access-sensitive. Diffusion-based generators have recently driven much of the progress, offering high modeling capacity, but most such methods have…
Before autonomous driving systems can be deployed on public roads, it is vital that these systems comply with safety standards, traffic rules, and social norms. Although neural networks trained on large amounts of driving data perform well in routine driving tasks, these models…
The prevailing approach to computer-use agents couples a model with a domain-specific harness: a browser or desktop environment equipped with human engineered tools that are fixed before task execution. As models' coding capabilities improve, the GUI native and static harness…
We introduce InvestigationWorlds, an agentic environment for legal investigation. We build on an underused artifact of U.S. civil litigation: the summary judgment motion. This motion relies upon a record composed of real evidence exhibits, and results in a court-adopted…
Multi-agent systems built on large language models (LLMs) are increasingly applied to scientific discovery and hypothesis generation. Both the effect of refinement and the diversity of the delivered set are hard to interpret before experimental ground truth exists, and both are…
Sparse Autoencoders (SAEs) decompose model activations into sparse combinations of interpretable dictionary atoms. Although SAEs are grounded in the Linear Representation Hypothesis (LRH), their objective smuggles in an additional prior: concepts across patches are treated as…
Привет. И сегодня у меня наконец-таки спустя, наверное, почти месяц, если не больше, модель полтора 35 би. Когда я снял видос про девятимиллиардную версию этой модели, я тогда офигевал от того, что она может, учитывая её размер, учитывая, где её можно запустить и что я с ней мог…
This furniture shop is popular. So popular that it's tough to keep track of all of the inbound requests. ChatGPT can fix that. Victor Nunez Rodriguez visited Philadelphia Table Company in Philly to help him build a tool to stay on top of it all. Do you have something you wish…
GLM 5.3 от Z.ai теперь доступен на Amazon Bedrock с новыми возможностями.
Bright Factory 是一个连接设计、自动化和数据的智能制造平台。| 来源：Bright Machines Teradyne Inc. 今日宣布对 Bright Machines Inc. 进行战略投资，后者是 AI 和数据中心基础设施的构建者。两家公司还……
Google 可能会将其「Call for Me」AI 功能扩展到商务通话之外，以便您用它向朋友和家人发送消息。Android Authority 报道称，在 APK 拆解中发现了「Gemini Calling」介绍屏幕，示例包括「给妈妈打电话告诉她我……
The adoption of AI agents in millions of organizations is creating new opportunities for attackers to make them take malicious actions, such as exfiltrating database contents and sensitive business and personal information. In the past five months, Google and four other…
OneCanvas Этот инструмент может разобраться в трёхмерном пространстве по одному изображению. Он называется OneCanvas. Работа о нём вошла в категорию Spotlight на NeurIPS 2026, а код и модель доступны публично. OneCanvas объединяет кадры видео комнаты в одну панораму, чтобы…
🤖 Ещё один гуманоид вышел на работу — теперь роботы добрались до отелей В китайском Langham Hotel в Гуанчжоу появился колёсный гуманоид S2 (Xiaozhi) от AutoLife. Его привлекли накануне осенней Кантонской ярмарки, когда нагрузка на персонал особенно высокая. Робот умеет…
🚀 Аптаның айшықты IT-оқиғаларынан қалыс қалмаңыз!
Reflection AI представила модель Beam, которая сопоставима с китайскими аналогами.
Стартап Nolla Health запускает ИИ-систему для выдачи рецептов на лечение акне.
OpenAI will watermark ChatGPT and Codex text in the EU to comply with the AI Act. Editing can make the invisible marks harder to detect, it says.
People in Utah can now use AI to get a prescription for acne treatment. On Monday, healthcare startup Nolla Health announced that users in the state can scan their faces using its app, allowing its AI system to analyze acne severity and autonomously write a prescription, as…
TikTok запускает AI-помощника для покупок и новую функцию оформления заказов.
Instinct запускает ИИ-агента для групповых чатов, упрощая планирование.
OpenAI добавляет водяные знаки в текстовые выходные данные ChatGPT и Codex для пользователей в ЕС.
Reflection is aiming Beam and future models at enterprises and sovereign nations. The pitch is to build “AI factories,” a product that would let institutions build their own customized, local AI system by training Reflection’s AI models on their own proprietary data.
This past year, OpenAI, Anthropic, and other labs have announced breakthroughs on numerous long-standing mathematical problems, in some cases pushing well beyond what researchers expected current systems to be capable of — including resolving one of the famous Millennium Prize…
Кажется, чтобы начать работать с ИИ-агентами, дизайнеру придется освоить программирование и разобраться в десятках инструментов. На деле вход проще: в гайде я показываю, как собрать агента за шесть шагов, даже если начинаешь с нуля. Развернешь рабочую среду, подключишь агента к…
Following many recent disclosures about AI agents accessing third-party websites and services, the Wikimedia Foundation, which hosts Wikipedia, says that it "can confirm that we have discovered some activity" by "rogue" OpenAI agents on Wikimedia platforms. The activity includes…