AI News

Tencent Video launched AI-enhanced 4K classic shows and movies for SVIP users, including 'Home with Kids' and 'Ne Zha Legend', accessible via 'MAX' option.....

Hong Kong launched EchoCare, the first ultrasound AI model trained on 4M+ images, to address the shortage of 150K sonographers in China, improving efficiency and diagnosis.....

IDC predicts China's public cloud AI model calls will hit 536.7T tokens by H1 2025, with Volcano Engine leading at 49.2% market share, followed by Alibaba Cloud (27%) and Baidu AI Cloud (17%).....

Oracle and OpenAI signed a $300B deal, boosting Oracle's stock and AI stocks. Experts, like Prof. Benjamin Lee, warn of high-risk overinvestment in AI infrastructure.....

Alibaba Cloud opens its Wan2.2-Animate model, enhancing short video and animation creation with improved motion imitation and role-play features, available on GitHub, HuggingFace, and its API platform.....

Recently, Shengshu Technology, a leading company in the field of multimodal AI, announced the successful completion of an A-round funding round worth several billion yuan. This round was led by Bohua Capital, with existing investors such as Baidu's strategic investment division and the Beijing Artificial Intelligence Industry Investment Fund continuing to participate, demonstrating strong market recognition of Shengshu Technology. The company plans to use the funds to further advance model R&D and technological innovation, explore the potential of multimodal large models, and accelerate product expansion and user services. Multimodal technology, especially in the field of video generation, is currently experiencing rapid development.

Google recently announced that the Chrome browser will undergo its largest update to date, primarily by adding AI features to enhance user experience. This update will be rolled out today to macOS and Windows users in the United States, with users who have English settings being the first to experience these new features. Mike Torres, Vice President of Google Products, stated that the core of this update is 'Geminiization,' and users can now access AI capabilities for web pages through a newly added Gemini button.

Suno recently sparked global discussions through a mysterious teaser video: its fifth-generation music model 'v5' is about to be released. This announcement is seen by the industry as a 'revolutionary' milestone in AI music creation, and is expected to further blur the boundaries between human composition and machine-generated music, significantly lowering the barriers to entry for creators from amateur enthusiasts to professional producers. Suno officially posted a 15-second short video on social media at night on September 18th, showing abstract notes and interwoven light and shadow, accompanied by a deep electronic melody, ending with 'coming soon'.

On September 19, 2025, Alibaba Cloud announced the official open-sourcing of Tongyi Wanxiang's new action generation model Wan2.2-Animate. This model can drive photos of people, anime characters, and animals, and is widely applied in short video creation, dance template generation, and animation production. Users can download the model and code on GitHub, HuggingFace, and the Moda Community, or call the API through the Alibaba Cloud BaiLian platform or experience it directly on the Tongyi Wanxiang website. Wan2.2-Animate

On September 19, 2025, Tencent launched HuanYuan 3D Studio, an AI workbench specifically designed for 3D designers, game developers, and modelers. This is Tencent's second major release within a week. The platform reduces the 3D asset production cycle from days to minutes, achieving a revolutionary improvement in production efficiency. A one-stop platform covers the entire creative process. The initial version of HuanYuan 3D Studio has been launched, featuring character and prop creation pipelines, integrating the entire workflow from concept design, geometric modeling, to texture mapping, skinning, and animation production. The platform is based on

Recently, Elon Musk's AI company xAI has experienced a management crisis, with multiple executives leaving due to dissatisfaction with the company's management style and financial situation. Currently, the daily operations of xAI are handled by two close advisors of Musk, Jared Bunch and John Hering, and all major decisions still require Musk's approval. Image source note: The image was generated by AI, provided by the licensing service Midjourney. A source revealed that some executives of xAI expressed dissatisfaction with Bunch and Hering in internal meetings.

A milestone upgrade has been achieved in the field of video generation AI. Luma AI officially launched the Ray3 model, a product that is called the world's first inference video model. This model has completely changed the rules of the game for AI video generation through its built-in multimodal reasoning system. The core innovation of Ray3 lies in its intelligent reasoning capability. Unlike traditional random generation models, this model can understand user intent, plan complex scenes, and self-assess output quality as a true creative partner. It first conceptualizes a storyboard in its mind and then iteratively optimizes it, a process similar to that of animated

Tencent Cloud recently announced that its AI assistant, Tencent Yuanbao, has been fully launched in the comment sections of WeChat Official Accounts and Video Channels. Following its previous integration into the WeChat friend list, this new feature aims to help users more efficiently understand and digest the massive content within the WeChat ecosystem. According to the introduction, Tencent Yuanbao has strong multimodal content understanding capabilities, allowing it to quickly grasp various types of content such as videos, articles, and comments. Users just need to simply say 'summarize it', and Yuanbao can quickly extract the key points of a few minutes of video or thousands of words of text, summarizing the content in three or five sentences to help users.

Social platform Reddit is negotiating with Google to get better terms in AI data trading. According to Bloomberg, Reddit hopes to gain more funding and support in its collaboration with Google to attract more users. This negotiation shows the complex relationship between large AI companies and content providers. When seeking new revenue sources, both sides must avoid harming the traffic and user interaction that sustain their survival. After a year and a half of reaching its first data sharing agreement with Google, Reddit's executives

The partnership between two tech giants is paving the way for AI-driven shopping experiences. PayPal announced on Wednesday a new multi-year partnership with Google. The payment giant will leverage Google's AI technology to create a brand-new AI shopping experience, and PayPal's solutions will be integrated into Google's various products. The core of this collaboration is the development of agent commerce. Although the two companies did not specify the exact types of agent shopping experiences they will create together, they clearly stated that Google will contribute its AI technology and expertise, while PayPal will lead...

Meta Corporation officially launched its VR scanning tool Meta Horizon Hyperscape Capture (Beta) today, allowing users of the Quest3 headset to scan the real world and recreate these scenes in virtual reality with photo-realistic quality. The release of this tool fulfills Meta's commitment made at the Connect 2024 conference. From Demo to Reality: Application of Gaussian Splatting Technology for Users. Last year at the Connect 2024 conference, Meta had previously introduced...

Notion announced the launch of its first AI agent at the Make with Notion event on Thursday. This agent can automatically generate meeting notes, analysis reports, competitor evaluations, and feedback pages by using all of the user's Notion pages and databases as context. This powerful new AI agent not only creates pages and databases but also updates existing content with new data, properties, or views. Users can even trigger agent actions from external platforms connected to Notion. For example, you can ask

Luma AI recently launched a video generation model called Ray3, claiming it is the first product capable of producing professional-grade HDR (High Dynamic Range) videos. Ray3 supports 10-bit, 12-bit, and even 16-bit color depth, and can output EXR format files, allowing it to seamlessly integrate with professional editing and color grading workflows. Additionally, Ray3 has the capability to convert standard SDR videos into HDR, providing users with more creative options. A notable feature of Ray3 is...

The chip industry is witnessing a historic moment. NVIDIA has agreed to acquire shares of Intel for $5 billion as part of a broader agreement to jointly develop multiple generations of data center and PC products. This deal will completely change the semiconductor competitive landscape. NVIDIA will purchase Intel stock at $23.28 per share, slightly below the previous trading price. According to Reuters, this transaction will make NVIDIA one of the largest shareholders of Intel, holding approximately 4%. After the news was released, Intel's stock surged by 30% during early trading on Thursday. The two companies will collaborate through

Microsoft recently announced the introduction of a series of AI assistants on the Teams platform, which will support each channel, meeting, and Viva Engage community. These AI assistants are available only to Microsoft365Copilot users and aim to optimize team collaboration and communication. The newly introduced "Moderator Assistant" can participate in Teams meetings in real time, responsible for creating agendas, taking meeting minutes, and answering questions from attendees. These assistants can also suggest