AI News

NVIDIA invests an additional $2 billion in CoreWeave, acquiring Class A shares at $87.20 each. The partnership aims to build 'AI factories' to add over 5 GW of AI computing capacity by 2030, integrating NVIDIA's latest tech, including next-gen Rubin chips replacing Blackwell.....

Due to regulatory and safety controversies, Meta will suspend the AI character function for global minors starting in January 2026 to develop a new version with enhanced parental control. Minor or suspected minor users will be restricted, but the AI assistant function will retain age-appropriate protection.

Baidu Intelligent Cloud has significantly increased its AI-related revenue growth target for 2026 from 100% to 200%, aiming to consolidate and expand its leading position in the AI cloud market and strive for the number one spot in the industry. Data from 2025 shows that Baidu has won 109 large model-related projects among the main cloud vendors in China, demonstrating strong market performance.

OpenAI CEO Sam Altman states that AI is reshaping corporate hiring strategies, leading to a significant slowdown in hiring at OpenAI as AI boosts efficiency. He advises companies to avoid mass recruitment, as AI may handle most tasks in the future.....

Guangzhou Taidong Technology's self-developed 'Taiji' Q&A reasoning model topped the SuperCLUE commercial marketing leaderboard with 85.82 points, surpassing major tech firms and demonstrating strong AI capabilities in vertical fields.....

Weimob launches 'AI Fitting' solution, using proprietary models and industry knowledge to enable realistic virtual try-ons, addressing e-commerce challenges like decision-making and high return rates, and reshaping retail with technology.....

Research shows AI's impact on employment began before ChatGPT's release, with demand for AI-related jobs declining in early 2022, months earlier than public awareness.....

AI is accelerating the transformation of traditional industries, with applications in construction, traffic management, and low-altitude economy emerging as new growth areas. Shanghai Harbor's subsidiary has made progress in satellite power systems and new solar cell research.....

Sogou Input Method 20.0 launches with full AI integration, evolving from a tool to a smart assistant. It upgrades core interactions via large models, enhancing accuracy, speed, and intelligence in voice, typing, and translation.....

Tencent Sogou Input法 20.0.0纪念版全面AI升级，基于腾讯混元大模型，增强AI语音、翻译和打字功能，语音识别延迟降低40%，准确率98%，方言识别提升30%。....

Baidu's Wenxin APP launched a new beta test on January 27, introducing the industry's first 'multi-user, multi-Agent' AI group chat feature. This innovation breaks the traditional one-on-one interaction model, allowing multiple specialized AI agents (e.g., group chat assistants, health managers) to coexist and collaborate in a single chat, forming a multi-dimensional AI think tank. The AI assistants not only deeply understand context but also pos....

OpenAI launches ChatGPT ads at $60 per thousand impressions, triple Meta's average. It leverages AI dialogue's 'high-intent scenarios' to boost ad conversion, redefining advertising in the AI era.....

Ali Health's AI app 'Hydrogen Ion' introduces 'Dynamic Evidence Positioning' to combat AI misinformation by upgrading static citations to 'live evidence', enabling precise source tracking and real-time validation of timeliness, authority, and logical consistency through daily updates of global medical guidelines and literature.....

Hangzhou Internet Court dismissed a lawsuit against an AI platform for providing false university address information, ruling it had no subjective fault as it fulfilled reasonable duty of care.....

DeepSeek-OCR2 introduces a novel 'visual causal flow' concept, enabling dynamic processing of document layouts for improved logical recognition in complex documents.....

Moon's dark side releases open-source Kimi K2.5 model, excelling in vision, code, and general tasks with native multimodal design for complex execution. It lowers interaction barriers, allowing users to upload photos, screenshots, or screen recordings for direct logic understanding, such as completing front-end development tasks via screen recordings.....

The 'White Tiger-VTouch' dataset, with over 60,000 minutes of cross-modal visual-tactile data, addresses the gap in large-scale visuotactile resources, aiding global robotics developers in enhancing tactile perception and advancing embodied intelligence.....

Alibaba launches Qwen3-Max-Thinking, a flagship inference model with over a trillion parameters and 36T tokens of pre-training data. Trained via reinforcement learning, it excels in factual knowledge and mathematical reasoning, now available on Qwen PC and web platforms, with app support coming soon, allowing users to switch to enhanced reasoning with one click.....

Ant Group's Lingbo Tech open-sources LingBot-Depth, a high-precision spatial perception model using Orbbec binocular 3D camera data to enhance depth sensing and 3D understanding, aiding robotics and autonomous driving in achieving more accurate and reliable 3D vision.....

Gallup's latest survey reveals a stagnation in workplace AI adoption, with only a slight increase from 45% to 46% in Q4 2025. Despite initial interest, heavy users remain scarce, with 26% using it frequently and 12% daily.....