Welcome to the 【AI Daily】column! This is your daily guide to exploring the world of artificial intelligence. Every day, we present you with the hottest AI news, focusing on developers and helping you understand technology trends and innovative AI product applications.
New AI Products Learn More: https://top.aibase.com/
1. A Man Sentenced to Ten Months for Using AI to Write Erotica, Illegally Profiting Over 20,000 Yuan
The Daye City People's Court in Hubei Province recently ruled on a case involving the use of artificial intelligence to write and profit from erotic novels. Defendant Ke Mou was sentenced to ten months in prison and fined for producing and selling obscene materials. The case highlights the legal risks of AI technology in literary creation and emphasizes the importance of adhering to laws and regulations. The court found Ke Mou's actions to be a serious violation of the law, causing negative social impact. Despite his plea of guilty, he was punished according to the law.
【AiBase Summary:】
💰 Ke Mou used an AI program to write erotic novels, illegally profiting over 20,000 yuan.
⚖️ The court sentenced Ke Mou to ten months imprisonment and a fine of 5,000 yuan.
📖 AI technology in literary creation must adhere to laws and regulations to avoid illegal activities.
2. 360 Zhi Nao Team Successfully Replicates DeepSeek Reinforcement Learning Results, Releases Open-Source Model Light-R1-14B-DS
The 360 Zhi Nao team recently successfully replicated the reinforcement learning results of DeepSeek and released the open-source inference model Light-R1-14B-DS. This model achieves reinforcement learning effects at a 14B parameter scale, significantly improving mathematical reasoning capabilities and surpassing many 32B-level models. In mathematical competition tasks, Light-R1-14B-DS performed exceptionally well, particularly showing significant score improvements in the AIME test. This achievement marks significant progress for smaller-scale models in the field of reinforcement learning, providing valuable resources and data to advance AI reasoning capabilities.
【AiBase Summary:】
📈 Light-R1-14B-DS achieves reinforcement learning effects at a 14B parameter scale, surpassing the DeepSeek-R1-Distill series of models.
🏆 In the AIME24 and AIME25 tests, Light-R1-14B-DS improved scores by 4.3 and 10 points respectively, demonstrating outstanding performance.
🛠️ Employing progressive supervised fine-tuning and reinforcement learning, two innovative training methods, significantly enhances the model's mathematical reasoning capabilities.
Details: https://github.com/Qihoo360/Light-R1
3. Say Goodbye to "Silent Video" Embarrassment! ByteDance's AI Sound Effects Generation Model SeedFoley Launches on Jiemong, Generating Cinematic Sound Effects with One Click
ByteDance's SeedFoley sound effects generation model has revolutionized the sound effects experience in video creation. Using advanced AI technology, the model achieves high synchronization between video and sound effects. Users can easily add professional-grade sound effects to their videos with simple operations. SeedFoley can process videos of various lengths and accurately identify video content to generate matching sound effects, greatly enhancing the narrative effect and emotional expression of videos. Now, users can easily experience this innovative feature on the Jiemong platform to create more impactful audio-video works.
【AiBase Summary:】
🎤 The SeedFoley sound effects generation model uses AI technology to achieve perfect matching between video and sound effects, enhancing the video creation experience.
⚙️ The model uses a revolutionary end-to-end architecture that can intelligently analyze video content and generate professional-grade sound effects.
🎶 Users can generate sound effects with one click on the Jiemong platform, easily avoiding the embarrassment of silent videos and enhancing the narrative power of videos.
4. Giant Network Releases Industry's First DeepSeek Native Gameplay
On March 14th, Giant Network announced that its social deduction game, "Space Kill," has officially integrated the DeepSeek large model, introducing the innovative gameplay "Inner Ghost Challenge." This not only marks the first in-depth application of DeepSeek technology in gaming but also reshapes the competitive and interactive experience through AI-driven core gameplay. The new gameplay combines the confrontation between real players and AI players, using AI's powerful reasoning capabilities to enhance the game's challenge and interactivity, and redefine the intelligent boundaries of human-computer confrontation.
【AiBase Summary:】
🤖 The "Inner Ghost Challenge" gameplay in "Space Kill" combines the confrontation between real and AI players, enhancing the game's interactivity.
🧠 The introduction of the DeepSeek large model allows AI players to possess powerful logical reasoning capabilities, accurately analyzing loopholes in conversations.
⚔️ AI players generate intelligent strategies through reinforcement learning, enhancing the game's challenge and optimizing the reasoning and interaction experience between players.
5. Sesame Releases CSM Model: Real-Time Emotionally Customized AI Speech Synthesis Reaches New Heights
Sesame released its latest speech synthesis model, CSM, on March 13th, attracting considerable industry attention. Based on the advanced Transformer architecture, this model can generate natural and emotionally rich speech, allowing users to adjust tone and emotional characteristics in real time. Although currently primarily supporting English, CSM is considered a significant breakthrough in AI speech technology, potentially reshaping the standards of human-computer dialogue and driving innovation in education and entertainment.
【AiBase Summary:】
🎤 The CSM model uses an end-to-end multimodal learning architecture that can understand context and generate natural speech.
⚙️ Users can adjust tone, intonation, and emotion in real time, demonstrating high flexibility.
🌍 Currently, CSM primarily supports English and does not yet support Chinese, but future expansion is possible.
6. OpenAI Chat Playground Upgraded to Prompts Playground
OpenAI recently announced the upgrade of its popular Chat Playground, now renamed Prompts Playground. This update introduces a new design and enhanced features aimed at providing users with powerful tools for testing, comparing, and iterating prompts. The redesign integrates advanced tools, including web and file search, to improve user experience and creative flexibility. The core goal of Prompts Playground is to enable users to efficiently explore and optimize interactions with AI models, making it an invaluable resource for both casual users and professionals.
【AiBase Summary:】
🔍 The new platform integrates advanced tools such as web and file search, enhancing user experience.
🗂️ Users can save and share model configurations and tool settings, facilitating collaboration among developers and researchers.
🚀 Prompts Playground is officially launched, providing users with a convenient experimental platform and customized possibilities.
Details: https://platform.openai.com/playground/p/T4U6rwVjng3kKl4UjaPr2CCZ?mode=chat
7. Yan Zhijie, Head of Alibaba's Tongyi Lab Speech Team, Resigns
According to reports, Yan Zhijie, head of Alibaba's Tongyi Lab speech team, officially resigned on February 15th. He has extensive experience in the field of intelligent speech, having worked at the University of Science and Technology of China and Microsoft Research Asia, and holding several important positions at Alibaba. Yan Zhijie has made significant contributions to speech recognition and synthesis, particularly in the application development during the era of large models. Although his future plans are unclear, the industry generally believes he may be venturing into AI entrepreneurship.
【AiBase Summary:】
🔍 Yan Zhijie officially resigned on February 15th; he was the P10-level head of the speech team at Alibaba.
🎓 He has a strong academic background in intelligent speech, having worked at the University of Science and Technology of China and Microsoft Research.
🚀 As a core member of DAMO Academy, he was responsible for developing Tongyi Tingwu, Alibaba's first publicly tested large model application.
8. Beijing Adds 34 Newly Registered Generative AI Services, Including Xiaomi AI Search and Monica
Recently, Cyberspace Administration of Beijing released the latest announcement on the registration of generative artificial intelligence services in Beijing, indicating that, according to relevant regulations, generative AI applications with registered large model capabilities are allowed to go online and provide services. As of March 14th, Beijing has added 34 new generative AI services, bringing the cumulative total to 46, including products such as Xiaomi AI Search and Monica. This move will further promote the development and application of generative AI, while requiring online services to publicly display their online numbers in a prominent location to ensure transparency and compliance.
【AiBase Summary:】
🆕 As of March 14th, Beijing has added 34 new generative AI services, bringing the cumulative total to 46.
📱 Applications including Xiaomi AI Search, Xiaomi AI Writing, and Monica have completed registration.
🔍 Online services must publicly display their online numbers on the product page to keep users informed.
9. SoftBank Acquires Sharp's Old Factory, Collaborates with OpenAI to Transform It into an AI Intelligent Body Data Center
Japan's display industry is undergoing a major transformation. Sharp Corporation's Sakai Display Products Corporation has ceased production, marking the end of domestic large-size LCD panel manufacturing. SoftBank Group acquired Sharp's old factory, planning to transform it into an AI data center, expected to be operational in 2026, supporting the development of Japan's AI industry.
【AiBase Summary:】
🏢 SoftBank acquired Sharp's Sakai factory for approximately 100 billion yen, planning to transform it into a large-scale AI data center.
🤝 SoftBank is collaborating with OpenAI to promote the commercial application of AI intelligent body technology, focusing on customized training of enterprise customer data.
📈 Sharp also signed an agreement with KDDI, and KDDI will also build an AI data center to further support the development of Japan's AI industry.
