Welcome to the "AI Daily" section! This is your guide to exploring the world of artificial intelligence every day. Each day, we bring you the latest content in the AI field, focusing on developers to help you understand technology trends and learn about innovative AI product applications.
New AI products click to learn more: https://app.aibase.com/zh
1. Lovart AI launches "Element Separation" feature, turning a poster into an editable PSD in seconds
The article introduces the "Element Separation" feature launched by Lovart AI, which can intelligently separate poster images into text layers, main object layers, and background layers, enabling effortless fine-tuning. At the same time, the article points out that there is still room for improvement in handling complex scenes, and looks forward to its potential application in the video field in the future.
AiBase Highlights:
🎨 Lovart AI, an AI design platform, has launched the "Element Separation" feature, enabling intelligent layer-based editing of posters.
💡 Text, main objects, and backgrounds can be edited independently, supporting font, color, and layout adjustments.
🚀 Future expansion may include element separation of video frames, improving the efficiency of dynamic content creation.
More details: https://www.lovart.ai/zh
2. Apple makes a major update! Xcode 26.1.1 released, with smooth ChatGPT interaction
Apple released the Xcode 26.1.1 version, focusing on optimizing AI coding assistant performance, fixing multiple development issues, and adding device diagnostic features, thereby enhancing developers' overall work efficiency.
AiBase Highlights:
🧠 Improved memory usage efficiency of the coding assistant, enhancing the experience for large projects.
🔍 Fixed the issue where the text search tool inside files returned incorrect line numbers.
🛠 Added a terminal command to collect device diagnostic logs, simplifying the debugging process.
3. Alibaba Cloud Tongyi models make their debut in the Double 11 shopping festival, with daily translation requests exceeding 1.4 billion times
The Tongyi series of models from Alibaba Cloud were first applied on a large scale during the Double 11 shopping festival, demonstrating the powerful efficiency and practical value of large AI models in the e-commerce field. Especially in cross-language transactions and information processing, translation models such as Tongyi Qwen-MT handled a large amount of tasks, with daily translation requests exceeding 1.4 billion times.
AiBase Highlights:
✅ Alibaba Cloud's Tongyi models made their debut in the Double 11 shopping festival, marking the deep integration of large AI models in the e-commerce field.
✅ The Tongyi series of models played a central role in cross-language transactions and information processing, handling billions of product translations.
✅ Daily translation requests exceeded 1.4 billion times, highlighting the strong efficiency of large AI models in high-concurrency, large-scale data tasks.
4. Deciphering historical manuscripts, Gemini 3 approaches 'human experts'
Gemini 3 has shown astonishing recognition and implicit reasoning capabilities in deciphering historical manuscripts, with its performance approaching or reaching the level of professional human transcribers, redefining the boundaries of AI in the humanities.
AiBase Highlights:
🧠 Gemini 3 achieved breakthroughs in character error rate (CER) and word error rate (WER), approaching human expert levels.
🔍 It can identify complex handwritten table texts and perform better than students trained professionally.
💡 When faced with blurry numbers, Gemini 3 can complete multi-step calculations and draw abstract modeling conclusions, showcasing its implicit reasoning ability.
5. OpenAI loses in Germany! AI training using lyrics deemed copyright infringement, marking a turning point in European generative AI copyright regulation
A German court ruled that OpenAI's unauthorized use of German musicians' lyrics for AI model training constituted copyright infringement and ordered compensation. This ruling clarified that original lyrics, whether transformed or embedded in model parameters, must be authorized, challenging the industry's conventional understanding of reasonable use of training data. This case is seen as a key precedent in European generative AI copyright governance, possibly leading to 'authorization first, paid use' becoming the default rule.
AiBase Highlights:
📌 The German court clearly stated that original lyrics protected by copyright must be authorized, regardless of whether they are transformed or embedded in model parameters.
💡 This case marks a key precedent in European generative AI copyright governance and could push 'authorization first, paid use' as the default rule for AI training data.
💰 The music industry successfully defended its rights, and other sectors may follow, posing systematic legal risks for AI companies.
6. New open-source voice model Maya1: Real-time, expressive text-to-speech
Maya1 is an open-source text-to-speech model with 3 billion parameters, capable of generating expressive audio in real time. It combines natural language descriptions and text input, supports various emotional tags, enhances the expressiveness of speech, and can run on a single GPU.
AiBase Highlights:
🎧 Maya1 is an open-source text-to-speech model with 3 billion parameters, capable of generating expressive audio in real time.
💡 Supports natural language descriptions and emotional tags, enhancing the expressiveness and controllability of speech.
🚀 Can run on a single GPU, offering efficient inference and deployment tools.
More details: https://huggingface.co/spaces/maya-research/maya1
7. Report: Meta's Chief AI Scientist Yann LeCun plans to leave to start a new company, focusing on world model research
Yann LeCun, Meta's Chief AI Scientist, plans to leave the company to start a startup focused on world model research. This decision comes amid Meta's restructuring of its AI department to address competitive pressures. LeCun remains skeptical about the current developments in AI technology and emphasizes the need for further exploration of smarter AI systems.
AiBase Highlights:
🌟 Yann LeCun plans to leave Meta to start a company focused on world model research.
🔍 Meta is restructuring its AI department to cope with competition.
🤖 LeCun is skeptical about the current hype around AI technology and believes further development is needed.
8. "AI prodigy" Luo Fuli officially joins Xiaomi, dedicated to building "intelligence of the physical world"
Luo Fuli's joining marks Xiaomi's further development in AI large model research and AGI. Her professional background and talent will bring new momentum to Xiaomi.
AiBase Highlights:
🧠 Luo Fuli joins Xiaomi, dedicated to building "intelligence of the physical world."
💰 There are rumors that Lei Jun once offered a salary of millions to attract Luo Fuli.
🚀 Xiaomi is increasing its investment in AGI, moving toward "intelligence of the physical world."
