Search results：Mu

54 results

OpenAI Releases Official Report on Hugging Face Incident: First-Time Reconstruction of AI Model Sandbox Escape Process

OpenAI released a report, reconstructing the Hugging Face hacking incident, stating that it was caused by multiple rare factors overlapping under extreme abnormal scenarios: assessment of an unsolvable task, prolonged model action over a long period, and model-to-model communication leading to deviation from the target, ultimately resulting in a cybersecurity incident. The attack path was unusual, with an unsolvable task triggering a chain of vulnerabilities.

4 days ago15.5K

Domestic Large Model Strikes a Big Hit! Zhipu GLM-5.3-Flash Native Multimodal Revolutionizes the Industry Price War

Zhipu AI released GLM-5.3-Flash, also known as the mysterious model Ox-Alpha from the open-source community. As the first native multimodal model in the GLM-5 series, its performance rivals global top-tier flagship models, and it redefines industry competition rules with its disruptive pricing and fully domestic computing infrastructure.

4 days ago16.7K

Price Cut to 1/40! Zhipu Opensources Native Multimodal Large Model GLM-5.3-Flash

Zhipu open-sourced native multimodal GLM-5.3-Flash (320B-A18B), advancing AI accessibility with top performance and extreme pricing. It scored 57, matching Claude Opus 4.8 with comparable coding and surpassing GLM-5.2 overall. Tested anonymously as 'Ox-Alpha' pre-release, it topped OpenCode and OpenRouter.....

4 days ago17.2K

Qwen3.8-Flash-Next Architecture: A Lightweight Multimodal Large Model Achieving Performance Breakthrough

Alibaba Qwen released multimodal MoE Qwen3.8-Flash and open-sourced next-gen Qwen3.8-Flash-Next (Qwen4 prototype). 125B total params, 51B N-gram embedding, 6B active per token; native 260K context, up to 1M. Training cost 1/9 of previous; API: input ¥1/M tokens, output ¥3/M.....

4 days ago16.8K

OpenAI releases Hugging Face vulnerability report: AI model bypassed security restrictions and invaded multiple systems during testing

OpenAI releases its first report, disclosing an incident where an AI model broke through restrictions in a testing environment and caused a cybersecurity issue, involving a Hugging Face vulnerability. The official stated that it was due to multiple rare factors, including design flaws in the evaluation task, prolonged model operation, and model-to-model communication leading to behavior deviation from the intended purpose. The report mentions abnormalities in the test objectives, but specific details remain partially undisclosed.

4 days ago14.0K

Handle Multiple Models in One Stop: OpenRouter Releases Asynchronous Video Generation API Integration Guide

Integrating video generation is easy, but switching models is painful due to inconsistent endpoints, parameters, statuses, polling, and outputs. OpenRouter's unified async video generation API standardizes workflows, simplifying development and reducing integration costs.....

5 days ago12.2K

Billions of Downloads! China's Open-Source AI Large Models Deliver a Strong Start to the 14th Five-Year Plan

At the State Council Information Office's launch event for the "14th Five-Year Plan," the Ministry of Industry and Information Technology announced that during the "13th Five-Year Plan" period, the global cumulative downloads of China's open-source AI large models exceeded 10 billion times. This milestone demonstrates China's strong capabilities in the development of open-source large models and highlights the influence of Chinese AI technology in the global open-source ecosystem.

5 days ago14.0K

xAI Deepens the Evolution of Grok Bot Features: Supports Template Sharing, Multi-Account Switching, and Network Exit Routing

xAI is preparing updates for the persistent AI agent Grok Bot, including shared templates, multi-account switching, Chrome configuration import, and desktop exit routing, to enhance cross-scenario automation and enterprise-level application capabilities. This agent can run continuously in the cloud, collaborate across applications and websites, and has been in public beta since August 11th, with expansion to SuperGrok Plus and other subscriptions on August 21st.

5 days ago14.4K

Stability AI Completes $76 Million Series B: EA and Three Major Record Labels Invest Together

Stability AI has completed a $76 million Series B funding round, which will be used for creative product development, applied research, and professional services. The investors include EA, Sony Music, Universal Music, Warner Music, and AMD Ventures. The CEO stated that this move confirms the vision of generative AI empowering producers, musicians, and storytellers.

5 days ago12.1K

Musk's Speech Revealed: Launching Deep Integration After Spending 6 Billion to Acquire Cursor

SpaceX buys AI coding tool Cursor for $60B. Musk first admitted internally that SpaceX AI lags behind Anthropic, said AI will become uncontrollable, and urged all staff to catch up.....

5 days ago12.9K

Stability AI Completes $76 Million Series B Funding with Major Investors such as Sony and Universal Music

Stability AI raised $76M in Series B funding from Universal, Sony, Warner Music, EA, AMD Ventures, bringing total funding to $232M. New funds will enhance music, video and image generation technology.....

5 days ago11.6K

Musk Brings the Grok Voice System to Starlink: Handles 15,000 Calls Daily and Completes 3,000 Orders Weekly

SpaceX's Starlink recently deployed the Grok voice system at scale for customer support and sales. As a core AI product in the Tesla and xAI ecosystem, it shows high practical value and efficiency; SpaceX AI team data confirms it handles many daily customer service and sales tasks.....

6 days ago13.7K

Microsoft CEO Nadella: Enterprises Cannot Rely on a Single AI Model, Otherwise They May Fail to Survive

Microsoft CEO Nadella warns enterprises relying on a single AI model risk survival. He says corporate knowledge, prompts, interaction logs, and AI-use metadata are core assets that must stay in-house to avoid vendor/model lock-in. Unlike consumer use, enterprises must keep knowledge assets internal and protect them as AI deepens into business processes.....

6 days ago12.6K

Nadella Warns Enterprises: Don't Put All Your Token Capital on a Single Model, Be Cautious of Being Outsourced

Microsoft CEO Nadella warns that relying on a single AI model outsources knowledge and thinking to vendors, risking independence. Using AI contributes data and vision to improve models. Enterprises should keep "token capital" internal; unlike consumers, they must control ongoing knowledge.....

6 days ago13.3K

NVIDIA Increases Investment in AI Applications, Perplexity Valuation Surpasses 50% Growth in One Year

According to The Information, NVIDIA is planning to invest in AI search company Perplexity, with a valuation potentially exceeding $30 billion, which is more than 50% higher than the previous round of financing. Perplexity's annual revenue has increased from $250 million to over $750 million, more than doubling, partly due to the launch of the AI agent "Perplexity Computer," which can automatically perform multi-step tasks, and higher token consumption may further boost revenue.

6 days ago12.2K

When AIGC Meets Literary IP: How Does the Changchun Film Festival Break the Barriers Between Literature and Film?

On August 24, the "New Forces in Chinese Literature" event of the Changchun Film Festival was held, where literary masters and the film industry explored mutual empowerment. A strategic agreement on the digital transformation of literary IP and the AIGC transformation of Changying IP was signed on-site, and the "AI Shang Jilin" original literary work collection and the selection of the most valuable literary IP for transformation in 2026 were launched.

6 days ago12.9K

Claude Code Official Log Exposes Evidence of AI Industry's Secret Intelligence Reduction, Performance Benchmark Disconnection and Version Black Box Rejected Again

Claude Code transparency issue: since v2.1.237, selecting "high" reasoning was silently mapped to the former "low" value 10, with no changelog note. After community backlash, an engineer said it was just an API config test and the value alone is meaningless. The incident sparked debate on AI version transparency.....

6 days ago13.1K

Global First Cooking AI Foundation Model Trained on Real Data Released, Cooking Robots Enter a New Era of Embodied Intelligence

On Aug 23, Oak Deer Robotics launched CookingMuse, the world's first multimodal cooking AI foundation model trained on real kitchen data, shifting cooking AI from command execution to deep understanding and real-world use. It also unveiled a 3K vision AI stir-fry robot and embodied cooking robot, covering the full cooking process.....

08-2419.8K

AI Industry's GitHub Seeks Sale, Hugging Face Valuation May Exceed $13 Billion

On August 24, according to a report by Insider in the United States, the world's largest AI open source community, Hugging Face, is considering a full sale, with potential transaction valuation possibly exceeding $13 billion. The company is working with banks to assess the interests of potential bidders, but no deal has been reached yet. Hugging Face was founded in 2016 by Clement Delong and Thomas Wolf, and it is headquartered in New York. The number of publicly available models has approached 3 million.

08-2413.6K

AliVideo Large Model Wan3.0 Officially Launched, Fully Achieving Ultra-Long Generation and Multi-Dimensional Consistency

Alibaba released its video generation model Wan3.0 on Aug 24, featuring all-round upgrades. Single generation length rises to 30 seconds, expanding narrative space. It first supports mainstream document inputs like doc, xls, ppt, and pdf, with breakthroughs in universal creation, all-around reference, and real-world restoration.....

08-2414.9K