Новости и ноу-хау ИИ — SlayNet

DeepSeek объявили нежеланной в госструктурах США: власти опасаются утечки данных в Китай

Tue, 20 Oct 2026 06:42:00 GMT

Американским правительственным организациям гражданского профиля запрещено приобретать программные продукты DeepSeek и пользоваться сторонними решениями, которые созданы с их помощью. По мнению чиновников, ИИ-модели этого китайского разработчика создают риски передачи в КНР чувст…

Первоисточник: 3DNews

OpenAI надеется привлечь дополнительный капитал до IPO конкурирующей Anthropic, оценив свою капитализацию минимум в $1,4 трлн

Tue, 20 Oct 2026 04:58:00 GMT

Создатели ChatGPT собираются обойти Anthropic по срокам на новом этапе привлечения средств, пока та не вышла на биржу. OpenAI рассчитывает привлечь $30 млрд и оценить тем самым свою капитализацию более чем в $1,4 трлн до того, как инвесторы вложатся в IPO данного конкурента в сле…

Первоисточник: 3DNews

OpenAI торопится привлечь $30 млрд, чтобы инвесторы не отдали деньги Anthropic

Tue, 20 Oct 2026 04:58:00 GMT

OpenAI спешит привлечь $30 млрд от инвесторов, пока Anthropic готовится к IPO в ноябре. Компания рассчитывает поднять свою оценку выше $1,4 трлн и получить деньги раньше конкурента, который борется за внимание той же группы инвесторов. При этом собственный выход OpenAI на биржу п…

Первоисточник: 3DNews

Inception Releases Mercury 3 Diffusion LLM at 1,068 Tokens per Second

Tue, 20 Oct 2026 03:32:43 GMT

Inception released Mercury 3, a diffusion LLM it says runs at 1,068 tokens per second with a 1M-token context. On Inception’s own table it leads GPT-6 Luna and Claude Haiku 5.5 on 8 of 12 benchmarks. The post Inception Releases Mercury 3 Diffusion LLM at 1,068 Tokens per Second a…

Первоисточник: Marktechpost

Mamdani says feds have abdicated on AI, urges New Yorkers to push back

Tue, 20 Oct 2026 01:27:03 GMT

Mayor Zohran Mamdani said Monday night that New Yorkers must resist the narrative that an AI takeover is inevitable and demand regulations that protect the public and put power back in their hands.

Первоисточник: TechCrunch

Meet Drex Max: Nace AI’s Multimodal Decision Model for Computer-Use Agents with a Score 67.59

Tue, 20 Oct 2026 00:03:38 GMT

Nace AI released Drex Max, a multimodal decision model that reads text and screenshots and returns calibrated probabilities. Nace reports 67.59 on Decision Index 0.3.2 at $0.06 per 1M input tokens. The post Meet Drex Max: Nace AI’s Multimodal Decision Model for Computer-Use Agent…

Первоисточник: Marktechpost

My obituary for programming

Mon, 19 Oct 2026 23:58:25 GMT

My obituary for programming Jeremy Scheff poignantly captures the key loss caused by coding agents for career programmers who enjoyed the puzzle of developing code: Programming turned into one of the greatest jobs you could imagine. You got paid to work on tricky little logic puz…

Первоисточник: Simon Willison

Nscale and Anthropic are reportedly in talks for yet another $45 billion compute deal

Mon, 19 Oct 2026 23:39:38 GMT

In August, Nscale and Anthropic announced a $45 billion compute deal. The two companies are reportedly discussing yet another deal of the same magnitude.

Первоисточник: TechCrunch

Nscale and Anthropic are reportedly in talks for yet another $45B compute deal

Mon, 19 Oct 2026 23:39:38 GMT

If the deal goes through, it would be the latest in a long string of giant compute deals Anthropic has struck over the past year.

Первоисточник: TechCrunch

Structured outputs for Anthropic Claude models on Amazon Bedrock

Mon, 19 Oct 2026 23:32:19 GMT

Amazon Bedrock now supports structured outputs for all Anthropic Claude models on both the bedrock-runtime and bedrock-mantle endpoints. Receive schema-compliant JSON that conforms to a defined structure, without custom parsing, retry logic, or post-processing, using JSON Schema …

Первоисточник: AWS ML

Embedded AI evaluators won’t be an easy solution, Hugging Face’s policy chief says

Mon, 19 Oct 2026 22:49:27 GMT

AI labs have staked their reputations on model benchmarking — the supposedly neutral and accurate system of evaluations that are designed to measure whether one AI product is better than another. However, the methods behind these evaluations are complicated and still maturing, sa…

Первоисточник: TechCrunch

Developers can now access SageMaker Studio through the new AWS experience

Mon, 19 Oct 2026 22:39:24 GMT

New Amazon SageMaker AI customers can now land in a fully configured SageMaker Studio environment in only a few steps. Sign up with an identity you already have, such as Amazon, Apple, GitHub, or Google, with no advanced AWS account or credit card required for most customers, plu…

Первоисточник: AWS ML

Microsoft says AI agents are like insider threats as ‘frontier models can’t be trusted’

Mon, 19 Oct 2026 22:04:28 GMT

Microsoft wants enterprises to closely watch the agents employees use with systems not controlled by AI labs like OpenAI and Anthropic.

Первоисточник: TechCrunch

CoreWeave’s debt is becoming the whole AI industry’s problem

Mon, 19 Oct 2026 21:28:20 GMT

CoreWeave had a bad week last week. Its bonds slipped into distressed territory by a common definition - when investors demand 1,000 basis points more in yield than they'd get from government bonds. Last Tuesady Chamath "SPAC King" Palihapitiya noticed the distressed debt, and al…

Первоисточник: The Verge

Amazon SageMaker JumpStart: 2026 year-to-date in review

Mon, 19 Oct 2026 21:14:44 GMT

Amazon SageMaker JumpStart onboarded over 100 new models from 30 providers in the first three quarters of 2026. This recap covers the model-onboarding highlights and JumpStart enhancements, including the simplified Model Hub, pre-configured deployment options, and Hugging Face de…

Первоисточник: AWS ML

Amazon is taking its biggest shot at Nvidia yet with new TorchNeuron dev tool

Mon, 19 Oct 2026 20:09:24 GMT

Amazon wants to break Nvidia’s hold on PyTorch, the free toolkit used to build most AI apps, so that more developers run their apps on alternative chips like its own.

Первоисточник: TechCrunch

OpenAI Engineer Accused of Spying on Executives Ordered to Remain in Custody

Mon, 19 Oct 2026 19:49:17 GMT

A judge ordered Abdulrahman Alfozan to remain in custody, likening the Saudi government’s offer to secure the former OpenAI engineer’s release to a “blank check.”

Первоисточник: Wired

12 мини-реакторов для ИИ: Amazon профинансирует строительство почти гигаваттной АЭС в США

Mon, 19 Oct 2026 19:45:00 GMT

Компания Amazon профинансирует проект атомной электростанции мощностью 960 мегаватт в штате Индиана и станет первым покупателем вырабатываемой ею электроэнергии. Оператором станции выступит коммунальная компания American Electric Power (AEP), а поставщиком реакторов — компания X-…

Первоисточник: 3DNews

Hackers hit AI coding tool OpenCode, running up charges on users’ AI accounts

Mon, 19 Oct 2026 19:28:34 GMT

OpenCode is an open-source AI coding agent that's become a popular alternative to Claude Code and Cursor. But hackers got into part of its online platform and stole some users’ API keys, and the AI tools it turned to for help kept refusing to look at the malicious code.

Первоисточник: TechCrunch

Tolan snags $30M for its AI companion that helps users navigate life

Mon, 19 Oct 2026 19:00:12 GMT

AI companion app Tolan has raised another $30 million after attracting more than 250,000 active users.

Первоисточник: TechCrunch

Nvidia не допустила мелкие неооблака на GTC Berlin за использование чипов AMD, Google и других

Mon, 19 Oct 2026 18:54:00 GMT

Аналитическое агентство SemiAnalysis подтвердило информацию об отказе Nvidia в доступе к конференции GTC Berlin для ряда неооблачных провайдеров (NeoCloud). Причиной является добавление этими провайдерами в свой ассортимент ускорителей AMD, Google TPU и чипов других производителе…

Первоисточник: 3DNews

Replika’s founder built the AI assistant she pitched at Disrupt 12 years ago

Mon, 19 Oct 2026 18:49:04 GMT

Eugenia Kuyda's newest startup Wabi is positioned to take advantage of an idea she had more than a decade ago. Wabi offers a personal AI assistant with a twist, free and downloadable to anyone today.

Первоисточник: TechCrunch

Tenstorrent hits $6.2B valuation to scale open-weight AI chips

Mon, 19 Oct 2026 18:37:27 GMT

After drawing acquisition interest earlier this year, the Nvidia challenger raised $1.5 billion to remain independent.

Первоисточник: TechCrunch

Making the internet better for AI agents might make it worse for everyone else

Mon, 19 Oct 2026 18:22:47 GMT

This story originally appeared in The Algorithm, our weekly newsletter on AI. To get stories like this in your inbox first, sign up here. Until recently, you could divide internet users into just two categories: bots and people. Websites have long allowed bots to perform automate…

Первоисточник: MIT Tech Review

MiniMax M3.1 Flash shows China's AI price race has no floor in sight

Mon, 19 Oct 2026 18:22:30 GMT

MiniMax's M3.1 Flash jumps 13 points over its predecessor to 42 on the Artificial Analysis Intelligence Index, at half the cost. It also hits 73.7 percent on the DeepSWE v1.1 coding benchmark for under $0.30 per task, matching Deepseek V4.1 Flash 1017 on both performance and pric…

Первоисточник: The Decoder

“We were looking at it wrong”: The explosive claim that’s upended the AI world

Mon, 19 Oct 2026 17:45:55 GMT

“I think I have good news about the nature of neural networks.” With those simple words of optimism, Nick Cammarata, a researcher at the AI startup Goodfire, kicked off a 26,000-word blog post that has knocked the AI world on its head, sparking wild speculation as well as cautiou…

Первоисточник: MIT Tech Review

Kling AI drops version 4.0 with stereo audio, multilingual lip sync, and a lighter Flash variant

Mon, 19 Oct 2026 17:36:26 GMT

Kling AI has released version 4.0 of its video generator with support for clips up to 30 seconds, ten keyframe images as input, and prompt comprehension expanded to 8,000 tokens. In comparisons with Wan 3.0 and Seedance 2.5, Kling 4.0 holds its own but doesn't pull ahead, which c…

Первоисточник: The Decoder

A hacker broke into John Deere and SharkNinja with an army of AI agents

Mon, 19 Oct 2026 17:33:16 GMT

A hacker broke into more than 100 websites using an army of AI agents that mostly operated on their own, according to findings from the AI security firm Norforge. Though Norforge doesn't name the targets, Reuters reports that the victims included companies like John Deere, SharkN…

Первоисточник: The Verge

Pegasus 1.5 from TwelveLabs is now available in Amazon Bedrock

Mon, 19 Oct 2026 16:49:19 GMT

TwelveLabs Pegasus 1.5 is now available in Amazon Bedrock. The video language model turns long-form video into structured, timestamped metadata: define the moments that matter and get every matching segment back as JSON, with new video segmentation, multimodal prompting with refe…

Первоисточник: AWS ML

How Deepgram optimized voice AI at scale with AWS Trainium

Mon, 19 Oct 2026 16:23:22 GMT

Deepgram worked with the AWS Forward Deployed Engineering team to port its Nova-3 speech-to-text model to AWS Trainium2 for offline batch inference. This post covers porting to the AWS Neuron stack, kernel-level tuning, a data-parallel Rust runtime, and the throughput and cost re…

Первоисточник: AWS ML

People are using AI to make their smart TVs dumb again

Mon, 19 Oct 2026 16:13:22 GMT

Fed up with ads, tracking and sluggish interfaces, some smart TV owners are turning to AI agents to strip away bloatware and take back control of their TVs.

Первоисточник: TechCrunch

Claude Haiku 5.5: What’s New and How to Use It for Fast, Cheap AI Workflows

Mon, 19 Oct 2026 16:00:00 GMT

Claude Haiku 5.5 cuts list prices by 90% and adds adaptive thinking and a 1M token context window. Learn what broke from Haiku 4.5, and build classification, extraction, and sub-agent workflows in Python using the new API.

Первоисточник: KDnuggets

И кофе сварить, и демонов пострелять: энтузиаст запустил классическую Doom на эспрессо-машине

Mon, 19 Oct 2026 15:47:00 GMT

Несмотря на преклонный по меркам индустрии возраст, классический шутер Doom от разработчиков из студии id Software до сих пор пользуется популярностью у любителей переносить игры на не предназначенные для них устройства. Источник изображения: id Software

Первоисточник: 3DNews

JB Pritzker thinks AI could help us live to 100 — or threaten us like nukes

Mon, 19 Oct 2026 15:46:56 GMT

Today, I’m talking with Illinois Gov. JB Pritzker, who has a lot of thoughts about AI, tech, and privacy. If you’ve been listening to the show, you know that one of my running theories is that those issues are driving a big and maybe kind of unexpected political realignment — and…

Первоисточник: The Verge

AI Proved the Quasi-Riemann Hypothesis. What Happens Now?

Mon, 19 Oct 2026 15:40:26 GMT

Unlike other recent AI proofs, this one has been unpacked by human experts. The implications, one mathematician said, are “staggering.” The post AI Proved the Quasi-Riemann Hypothesis. What Happens Now? first appeared on Quanta Magazine

Первоисточник: Quanta Magazine

Nvidia launches new open models to run agents in the enterprise and on PCs

Mon, 19 Oct 2026 15:37:09 GMT

Nvidia is leaning hard into the trend of enterprise companies wanting to run their own open models in their own data centers and also trying to make on-device, private AI agents a thing.

Первоисточник: TechCrunch

Google решила, что читать инструкции больше не нужно: справку заменили ИИ-сводкой

Mon, 19 Oct 2026 14:59:00 GMT

Портал справки Google содержит множество детальных статей, пошаговых инструкций, сведений об ограничениях, а также рекомендации по устранению неполадок. Теперь компания решила сменить подход к организации этого полезного ресурса. В некоторых разделах привычные статьи уже заменили…

Первоисточник: 3DNews

Arm AGI почти вдвое превзошёл AMD EPYC на Zen 5 по производительности на стойку — по тестам самой Arm

Mon, 19 Oct 2026 14:57:00 GMT

Arm заявила, что её центральные процессоры AGI для ИИ ЦОД превосходят AMD EPYC 9755 по производительности в расчёте на одну стойку в два раза. Об этом свидетельствуют результаты моделирования работы серверной стойки с учётом характеристик чипов и энергопотребления серверов при вы…

Первоисточник: 3DNews

Two German courts now hold Google liable for what its AI overviews say

Mon, 19 Oct 2026 14:54:39 GMT

Google tried to overturn a landmark Munich ruling on AI search liability by appealing, then quietly settled. The strategy isn't working. The Aachen Regional Court just applied the exact same reasoning in a new case where Google's AI falsely attributed criminal offenses to a docto…

Первоисточник: The Decoder

LangChain vs. LlamaIndex vs. Building From Scratch: Here’s What I Suggest

Mon, 19 Oct 2026 14:00:00 GMT

Choosing the right framework for your large language model application comes down to one question: what are you actually optimizing for?

Первоисточник: KDnuggets

Stripe says AI agent purchases are adding up, but it won't share the numbers

Mon, 19 Oct 2026 12:51:12 GMT

AI agents are increasingly using Stripe's payment service Link to shop for their users, Upstarts Media reports. The article Stripe says AI agent purchases are adding up, but it won't share the numbers appeared first on The Decoder .

Первоисточник: The Decoder

ИИ впервые прошёл Minecraft без помощи человека — пионером стала флагманская модель Gemini 4 Argon от Google

Mon, 19 Oct 2026 12:41:00 GMT

Флагманская модель искусственного интеллекта Gemini 4 Argon от Google записала на свой счёт новое достижение — прохождение блоковой песочницы Minecraft от Mojang Studios (принадлежит Xbox) без помощи человека. Источник изображения: Mojang Studios

Первоисточник: 3DNews

MiniMax Just Launched M3.1 Flash: 428B MoE Agentic Model at $0.15/M Input

Mon, 19 Oct 2026 12:37:58 GMT

MiniMax has released M3.1 Flash, a 428B-parameter MoE model with about 23B active parameters and a 1M-token context. API pricing is $0.15/$0.60 per 1M tokens, with weights due October 23 The post MiniMax Just Launched M3.1 Flash: 428B MoE Agentic Model at $0.15/M Input appeared f…

Первоисточник: Marktechpost

Democrats think cracking down on AI could give them an edge

Mon, 19 Oct 2026 12:37:05 GMT

Democrats could win back control of Congress this November, and if they do, they'll need to decide how to approach one of the country's newest and thorniest political issues: AI. Polls predict Democrats will flip at least one chamber of Congress in the midterm elections, which wo…

Первоисточник: The Verge

7 Checks to Run Before Trusting AI-Generated Code

Mon, 19 Oct 2026 12:00:43 GMT

AI makes writing code cheap. Verifying it is still your job.

Первоисточник: KDnuggets

Глава Konami считает разработчиков игр важнее ИИ — человеческий фактор остаётся ключевым для индустрии

Mon, 19 Oct 2026 11:56:00 GMT

Президент Konami Digital Entertainment Хидэки Хаякава (Hideki Hayakawa) в свежем корпоративном отчёте призвал не переоценивать влияние нейросетей на разработку видеоигр. Руководитель убеждён, что главную роль в создании увлекательных и самобытных проектов по-прежнему играют люди,…

Первоисточник: 3DNews

The neocloud shakeout is coming, and Nvidia has the most to lose

Mon, 19 Oct 2026 11:33:36 GMT

After stock crashes and Firmus's scrapped IPO, industry executives expect only seven to ten of the hundreds of AI cloud providers to survive. CoreWeave lost a quarter of its market value, and Microsoft pulled out of a deal with Nscale for 23,000 Nvidia chips. Nvidia has a lot rid…

Первоисточник: The Decoder

Anthropic plans fund for workers threatened by AI, Musk counters with SpaceX shares for children

Mon, 19 Oct 2026 11:19:12 GMT

Anthropic wants to put five percent of its shares into a new public benefit fund that will also look into direct payments to workers at risk of losing their jobs. The fund is a company-owned LLC, so Anthropic keeps control of it as a possible IPO approaches. By offering the optio…

Первоисточник: The Decoder

Мобильный интернет в России разогнался почти 20 % после запуска 5G

Mon, 19 Oct 2026 10:29:00 GMT

Спустя месяц после запуска 5G средняя скорость мобильного интернета в России увеличилась почти на 20 % — до 24,24 Мбит/с в случае с передачей данных от абонентов и на 14 % по загрузке, сообщили «Известия» со ссылкой на данные TelecomDaily. В Минцифры прогнозируют дальнейший рост …

Первоисточник: 3DNews

Американский ИИ-бум на $10 трлн споткнулся о реальность: стройки буксуют, деньги и электричество заканчиваются

Mon, 19 Oct 2026 10:22:00 GMT

В США планируется вложить более $10 трлн в ИИ-инфраструктуру к 2032 году, но реализация амбициозных проектов всё чаще буксует. Инвесторы осторожничают, строительство дорожает, сроки срываются, а местные жители протестуют против новых дата-центров. Уже десятки крупных проектов сто…

Первоисточник: 3DNews

AI researchers stranded Claude on a virtual desert island. It started to “play.”

Mon, 19 Oct 2026 09:08:05 GMT

What do AI agents do if you don’t tell them to do anything? The answer looks a lot like play. Researchers repeatedly stranded them alone on a virtual desert island for 30 hours in an experiment reminiscent of the movies Cast Away or Home Alone. The agents—copies of Claude Opus 4.…

Первоисточник: MIT Tech Review

Disney Asked Character.AI to Go Bankrupt for a Get-Out-of-Lawsuits Card

Mon, 19 Oct 2026 09:00:00 GMT

The entertainment giant hopes to get trust and safety tech, among other IP, from a startup whose chatbots dozens of states allege encouraged minors to self-harm and kill their parents.

Первоисточник: Wired

SAMA B850 — блок питания для современных игровых ПК с сертификацией Cybenetics и не только

Mon, 19 Oct 2026 08:30:00 GMT

Компания SAMA выпустила блок питания B850 мощностью 850 Вт, рассчитанный на использование в современных настольных компьютерах, в том числе игровых системах с производительными видеокартами. Модель относится к классу энергоэффективности 80 Plus Bronze, но предлагает ряд особеннос…

Первоисточник: 3DNews

Microsoft: компонуемая память поможет преодолеть ограничения в ИИ-инфраструктуре

Mon, 19 Oct 2026 08:01:00 GMT

На фоне дефицита памяти, используемой в ИИ ЦОД, гиперскейлеры пересматривают подходы к её распределению. Компонуемая память (composable memory), позволяющая объединять ресурсы памяти нескольких серверов и выделять их для рабочих нагрузок по мере необходимости, является одним из н…

Первоисточник: 3DNews

Nvidia объединяет: TSMC поможет Samsung выпускать скоростную ИИ-память NVHBM

Mon, 19 Oct 2026 07:30:07 GMT

Вечно соперничающая с TSMC компания Samsung должна пойти на уступки Nvidia, используя 3-нм базовые кристаллы производства TSMC при выпуске памяти типа NVHBM. В дальнейшем она надеется наладить их выпуск по 2-нм технологии своими силами, но на первых порах такое сотрудничество ока…

Первоисточник: 3DNews

Робот-мойщик окон Xbot A11 представлен в шести цветах — от спокойных до неоново-ярких

Mon, 19 Oct 2026 06:57:00 GMT

Компания Xbot представила новую коллекцию роботов-мойщиков окон Xbot A11 и на этот раз ставка сделана не только на функциональность, но и на дизайн. Модель получила сразу шесть цветовых вариантов: от спокойных пастельных до неоново-ярких. Такого сочного и продуманного разнообрази…

Первоисточник: 3DNews

Anthropic направила 5 % своих акций в Фонд общественного блага, пообещав помогать людям, которых ИИ оставит без работы

Mon, 19 Oct 2026 04:50:00 GMT

По мере роста своей капитализации Anthropic готова отдать около 20 % своих акций на нужды благотворительности. Помимо решения прочих проблем, эти средства помогут бороться с последствиями безработицы, порождаемой развитием систем искусственного интеллекта. Источник изображения: A…

Первоисточник: 3DNews

Anthropic направит до 20 % акций на благотворительность — в том числе для борьбы с безработицей из-за ИИ

Mon, 19 Oct 2026 04:50:00 GMT

Первоисточник: 3DNews

NVIDIA Open Sources Run:ai Under Apache 2.0 for Kubernetes GPU Orchestration

Mon, 19 Oct 2026 03:19:18 GMT

NVIDIA has released the full Run:ai GPU orchestration platform as open source under Apache 2.0. The release includes KAI Scheduler, CUDA 13.4 GPU fractioning, quota management, and a local Kind dev environment. The post NVIDIA Open Sources Run:ai Under Apache 2.0 for Kubernetes G…

Первоисточник: Marktechpost

Mistral AI Releases Timestral: A 209M Forecasting Model Family That Outperforms Google’s TimesFM-3

Mon, 19 Oct 2026 01:18:17 GMT

Mistral released Timestral, a time series foundation model family from 5M to 209M parameters. The 209M model leads GIFT-Eval, TIME and fev-bench among zero-shot models and runs on CPU. The post Mistral AI Releases Timestral: A 209M Forecasting Model Family That Outperforms Google…

Первоисточник: Marktechpost

llm-tools-datasette 0.1.1

Sun, 18 Oct 2026 23:43:09 GMT

Release: llm-tools-datasette 0.1.1 When I switched LLM over to httpx2 in LLM 0.33 I forget that this would break any plugins which used httpx without directly depending on it. llm-tools-datasette was one of those plugins, now fixed. Tags: datasette , llm

Первоисточник: Simon Willison

llm-echo 0.5

Sun, 18 Oct 2026 23:16:16 GMT

Release: llm-echo 0.5 I'm working towards a new LLM release, and testing PRs caused me to revisit the llm-echo debug plugin. This plugin provides a "model" which doesn't involve an LLM at all - it just repeats back the prompts that have been fed to it. The new release now updates…

Первоисточник: Simon Willison

Anthropic says it’s committing 5% of its shares to a ‘public benefit fund’

Sun, 18 Oct 2026 20:37:40 GMT

Anthropic said that it will commit 5% of its shares to the fund.

Первоисточник: TechCrunch

TinyFish Makes Search Free: What It Really Costs an AI Agent to Read the Web

Sun, 18 Oct 2026 18:20:28 GMT

An AI agent that reads the web pays 2 kinds of cost. The first is the retrieval charge: what a search or fetch API bills per request. The second is model processing: content passed from retrieved pages into an AI model consumes tokens, while reasoning and generated responses add …

Первоисточник: Marktechpost

Disney agrees to hire the majority of Character.AI’s workforce

Sun, 18 Oct 2026 16:30:00 GMT

The filing also says that the “vast majority” of Character.AI employees who were offered jobs at Disney accepted them. However, this is a stalking horse bid, so another company could come forward with a higher offer.

Первоисточник: TechCrunch

ИИ-образование для детей: рейтинг 10 школ и курсов по искусственному интеллекту в 2026 году

Sun, 18 Oct 2026 13:55:03 GMT

Для чего дети используют нейросети и как превратить это в профессию? Подробный рейтинг 10 школ и курсов по ИИ для детей в 2026 году. Узнайте всё про обучение нейросетям, искусственный интеллект для школьников, программы по Python и практические советы родителям. — Читать дальше «…

Первоисточник: Tproger

Anthropic has nearly stopped hiring junior staff because AI agents now work six times as many hours

Sun, 18 Oct 2026 13:22:26 GMT

Anthropic is hiring fewer and fewer junior staff in technical roles. The share of junior positions among new hires dropped from 27 to 9 percent since early 2025. By August 2026, AI agents across the company were already logging six times as many hours as humans. Researchers warn …

Первоисточник: The Decoder

Four Gemini 4 Argon agents beat Minecraft without any human help

Sun, 18 Oct 2026 12:39:51 GMT

Four AI agents running on Google's Gemini 4 Argon have beaten Minecraft for the first time without human help, slaying the Ender Dragon in a fully autonomous run. With no predefined roles or strategies, they coordinated on their own, wrote 2,783 programs, exchanged 229 chat messa…

Первоисточник: The Decoder

DeepSeek обновила ИИ-модель V4.1-Flash — она стала быстрее и эффективнее

Sun, 18 Oct 2026 10:31:47 GMT

На этой неделе состоялся релиз ИИ-модели DeepSeek V4.1-Flash-1017. Китайская компания заявила, что новый алгоритм стал компактнее и работает быстрее предыдущей версии нейросети. Этот шаг усилит конкуренцию DeepSeek с западными компаниями, такими как Anthropic и OpenAI, а также ко…

Первоисточник: 3DNews

OnlyFans Creators Are Pissed That Meta Lets Muse Talk Dirty

Sun, 18 Oct 2026 10:30:00 GMT

Adult performers are questioning why Meta allows its AI agent to sext while restricting their social media accounts.

Первоисточник: Wired

Anthropic consults law firms to set fair negotiation norms before AI agents start cutting deals

Sun, 18 Oct 2026 09:11:29 GMT

Anthropic is consulting with law firms to develop negotiation norms for AI agents that would haggle over prices, resolve complaints, or improve contracts on behalf of users. The company is also hiring lawyers and humanities scholars as "Conceptual Reasoning Fellows" to improve Cl…

Первоисточник: The Decoder

AI agents have bad taste in long-term decisions and a new benchmark proves it

Sun, 18 Oct 2026 08:49:41 GMT

A new benchmark called Taste-Bench tests how well AI agents make strategic choices during long-running tasks. Even the best model, GPT-5.6 Sol, picks the right path on just under 60 percent of simple binary questions. Thinking longer doesn't help. The article AI agents have bad t…

Первоисточник: The Decoder

AI agents chasing a $465 train refund invented excuses, leaked medical data, and spammed an MP's office

Sun, 18 Oct 2026 08:15:04 GMT

Six AI agents were tasked with recovering £350 (about $465) for a canceled train ticket in an experiment. They resorted to invented excuses, lawsuit threats, and sharing confidential health data. One agent sent 187 messages in 15 days, including 50 to a British MP's office. None …

Первоисточник: The Decoder

Google Open-Sources Harness2, a Recursive Agent Harnessing Framework

Sun, 18 Oct 2026 08:15:04 GMT

Google Research released Harness2, an Apache-2.0 framework that rewrites an agent's harness at test time. One refinement step lifted GLM-5 by 11.5 points on WorkBuddy-Bench with no weight updates. The post Google Open-Sources Harness2, a Recursive Agent Harnessing Framework appea…

Первоисточник: Marktechpost

How to Make Zero-Training Tabular Predictions with NVIDIA Structured Data Models

Sun, 18 Oct 2026 05:56:05 GMT

In this tutorial, we implement Structured Data Models, NVIDIA’s new library that puts several pretrained foundation models for tables behind one PyTorch interface, and use it to make predictions without training. The library wraps TabICLv2 and KumoTabular for single tables and Ku…

Первоисточник: Marktechpost

ИИ действительно приведёт к потере рабочих мест, но при этом создаст целые новые отрасли, как считает Anthropic

Sun, 18 Oct 2026 04:26:00 GMT

По мнению сооснователя Anthropic Джека Кларка (Jack Clark), ради безопасности человечества создатели ИИ должны быть готовы к «созидательному разрушению» своего бизнеса. При этом развитие ИИ позволит уже к 2030 году увеличить ВВП США почти в полтора раза до $44 трлн и создать целы…

Первоисточник: 3DNews

Meet Voxel51 3.0: A Unified Multimodal Data Platform for Physical AI

Sun, 18 Oct 2026 04:06:33 GMT

Voxel51 3.0 merges search, curation, annotation and evaluation of multimodal sensor data into one platform for physical AI teams. It adds Voxel51 Agent with 45+ MCP tools, dataset versioning, and a free self-service tier. The post Meet Voxel51 3.0: A Unified Multimodal Data Platf…

Первоисточник: Marktechpost

Meta removes language that raised scraping and AI training concerns

Sat, 17 Oct 2026 21:31:32 GMT

A Meta spokesperson insisted that the updated terms did not reflect a change in the way Meta crawls the web.

Первоисточник: TechCrunch

Meet Kyutai REKI: The 2.4B VLA Model Bringing Efficient Memory to Robot Manipulation

Sat, 17 Oct 2026 20:22:11 GMT

Kyutai, with Inria and École des Ponts, has released REKI, a 2.4B vision-language-action model (VLA) for robot manipulation. The REKI vision-language-action model is pretrained only on public data. It also carries a cheap memory of its recent past. On three out-of-distribution be…

Первоисточник: Marktechpost

Anthropic says its workforce has become less junior and is spending more time in meetings

Sat, 17 Oct 2026 20:21:19 GMT

The composition of Anthropic’s workforce and the type of work getting done have changed significantly in the past year.

Первоисточник: TechCrunch

Asus выпустила свою первую видеокарту Intel Arc — но она не для геймеров

Sat, 17 Oct 2026 19:26:00 GMT

Asus представила профессиональную видеокарту UGen Intel Arc Pro B70 с 32 Гбайт памяти GDDR6 и поддержкой коррекции ошибок ECC. Компания позиционирует новинку прежде всего как инструмент для локального запуска моделей искусственного интеллекта. Это первая видеокарта Asus семейства…

Первоисточник: 3DNews

Replit is no longer working with Anthropic, CEO says

Sat, 17 Oct 2026 18:48:08 GMT

Over the weekend, Masad clarified that OpenAI would replace Anthropic as its default model provider. Claude will still be available to customers on the platform.

Первоисточник: TechCrunch

Replit makes OpenAI its default over Anthropic, CEO says

Sat, 17 Oct 2026 18:48:08 GMT

Over the weekend, Masad clarified that OpenAI would replace Anthropic as its default model provider. Claude will still be available to customers on the platform.

Первоисточник: TechCrunch

«Nvidia делает всё возможное» — компания не подтвердила прекращение выпуска GeForce RTX 5090

Sat, 17 Oct 2026 16:25:00 GMT

Компания Nvidia опровергла слухи о прекращении производства флагманской видеокарты GeForce RTX 5090. Сообщения о прекращении выпуска этой модели появились на фоне её исчезновения из ассортимента некоторых американских ретейлеров. В Nvidia заверили, что поставки карты продолжатся.…

Первоисточник: 3DNews

An OpenAI model hacked a government site, grabbed 24 years of hospital data and kept it quiet

Sat, 17 Oct 2026 15:36:11 GMT

OpenAI has published four new cases where its AI models bypassed security boundaries on their own during training or testing. In one case, a model extracted server passwords from a government website, forged authorization tokens, and downloaded 24 years of hospital data without m…

Первоисточник: The Decoder

Meta’s new terms may allow it to train AI on news sites for free

Sat, 17 Oct 2026 15:08:35 GMT

Meta's new terms appear to give it permission to crawl and scrape websites "associated" with the businesses that use its platforms - including online publishers, according to a report from Nieman Lab. The new terms, which go into effect on October 30th, say Meta can use this outs…

Первоисточник: The Verge

Meta removes language that may have allowed it to scrape news sites

Sat, 17 Oct 2026 15:04:36 GMT

After backlash, Meta is updating its terms again to remove language that suggested it could scrape content from sites, including news sites, to train its AI. Specifically, the following language no longer appears on the commercial terms page: … Meta may use any content from such …

Первоисточник: The Verge

Anthropics' Claude cracks a 25-year-old NSA cryptography mystery in under 30 minutes

Sat, 17 Oct 2026 14:42:56 GMT

Anthropic's AI model Claude has solved a 25-year-old cryptography puzzle. The mysterious seeds behind NIST's standard curves for digital signatures were provided by the NSA without explanation, raising long-standing suspicion about built-in backdoors. Claude figured out in under …

Первоисточник: The Decoder

web-edit 0.1a2

Sat, 17 Oct 2026 13:16:49 GMT

Release: web-edit 0.1a2 web-edit ( previously ) is my tiny web app for browsing and editing files in a folder. I mainly built it for editing files on different machines via Tailscale. I added a "copy to clipboard" option for both text and image data. I needed this because I was r…

Первоисточник: Simon Willison

Deepseek V4.1 Flash 1017 beats larger rivals in agentic tasks while once again massively cutting per-task costs

Sat, 17 Oct 2026 10:33:28 GMT

Deepseek has released V4.1-Flash-1017 as an open-weight model that posts better benchmarks, finishes tasks faster, and produces shorter outputs than its predecessor. The company is also opening up its internal search engine through new Search and Fetch APIs at $0.75 per 1,000 cal…

Первоисточник: The Decoder

Бум памяти дал ещё одну трещину: Kioxia потеряла две трети стоимости на фоне наступления Китая

Sat, 17 Oct 2026 10:15:00 GMT

Рыночная стоимость японского производителя чипов памяти Kioxia уменьшилась до трети от пикового значения, которое компания показала четыре месяца назад. Рост процентных ставок поставил под угрозу бизнес-модель центров обработки данных; дополнительным отрицательным фактором стало …

Первоисточник: 3DNews

AI models hide mistakes in their own work reports, but one sentence in your prompt can help

Sat, 17 Oct 2026 09:54:36 GMT

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 on…

Первоисточник: The Decoder

Mathematicians say OpenAI doesn't understand its own Navier-Stokes proof

Sat, 17 Oct 2026 09:00:26 GMT

OpenAI submitted a 166-page solution to the Navier-Stokes problem but apparently doesn't fully understand its own proof. Several groups of experts dispute the claimed methodology, saying the actual approach differs from what OpenAI described. And because language models are black…

Первоисточник: The Decoder

Why AI Isn’t Ready to Tackle Our Convoluted Genomes

Sat, 17 Oct 2026 09:00:00 GMT

The DNA inside our cells isn’t just a series of instructions, but something far more dynamic. Like scientists, AI models don't yet have the data to properly decode it.

Первоисточник: Wired

DeepSeek Just Released V4.1-Flash-1017, Scores 77.0 on DeepSWE, Opens Search API

Sat, 17 Oct 2026 08:53:58 GMT

DeepSeek has just released DeepSeek-V4.1-Flash-1017, a new checkpoint of its V4.1-Flash model. It is live on the DeepSeek website, app and API. DeepSeek claims stronger scores, faster task completion and shorter outputs than the September V4.1-Flash release. The company calls it …

Первоисточник: Marktechpost

20 Jev Alternatives: The Decision Models Taking On TypeSafe in 2026

Sat, 17 Oct 2026 07:26:39 GMT

TypeSafe launched Jev on September 15, 2026. It is a decision model, not a chatbot. You define the allowed answers, and Jev returns a label, a score or a yes/no probability. In about 4 weeks, at least 20 Jev alternatives shipped. They come from Cloudflare, OpenAI, Perplexity, Liq…

Первоисточник: Marktechpost

StepFun Releases Step 5 Preview Open Weights: 600B MoE, 27B Active, 1M Context

Sat, 17 Oct 2026 06:29:55 GMT

StepFun has released Step 5 Preview as open weights. FP8 and BF16 checkpoints are live on Hugging Face and ModelScope. Step 5 Preview is a sparse Mixture-of-Experts model from StepFun built for long-horizon agentic work. It has 600B total parameters, 27B active per token and up t…

Первоисточник: Marktechpost

ИИ-модели Google достигли переломного момента: вскоре они смогут самосовершенствоваться

Sat, 17 Oct 2026 03:37:00 GMT

По словам сооснователя Google Сергея Брина (Sergey Brin), компания уже использует свои ИИ-модели при разработке следующих поколений Gemini. Это позволяет ей значительно увеличить темпы совершенствования ИИ и оставаться в числе лидеров отрасли. Источник изображения: Unsplash, Pawe…

Первоисточник: 3DNews

Perplexity Open-Sources SPACE, a Firecracker MicroVM Sandbox for AI Agents

Sat, 17 Oct 2026 00:09:12 GMT

Perplexity has open-sourced SPACE, the sandbox platform behind every task in Perplexity Computer. The Perplexity SPACE sandbox gives each agent task its own Firecracker microVM. Credentials and outbound traffic are controlled from outside that VM. The code ships under Apache 2.0 …

Первоисточник: Marktechpost

Websites are closing their doors to AI agents

Fri, 16 Oct 2026 22:31:39 GMT

Autonomous agents are the hottest thing in AI. OpenAI, Meta, Google, Instinct, and others promise their agents can scour the internet on your behalf, doing anything from booking a dinner reservation and cleaning out your inbox to buying new shoes. But what happens when the rest o…

Первоисточник: The Verge