2025年5月15日是全国第七个"515全国投资者保护宣传日"，为了更好地发挥主流媒体在投资者保护中的 积极作用，《证券日报》社有限责任公司举办了第七届"5 15—5 19中小投资者保护宣传周"公益活动。 近期，北京神州泰岳（300002）软件股份有限公司（以下简称"神州泰岳"）董事会秘书、资本运营中心 总经理、首席品牌官张开彦接受《证券日报》记者专访，并分享了神州泰岳在践行投资者保护理念过程 中的积极举措。 "手游行业迭代快，用户偏好流动性强，这反而要求企业持续沉淀、构筑起自己的护城河。"张开彦表 示，神州泰岳的三个"坚持"，就是要在尊重产业发展规律的同时，持续做自己擅长的事情，不断提高产 品的成功率。 此外，神州泰岳还积极回应投资者询问，对投资者关心的二级市场热点事项客观真实地表述。"不蹭热 点，不搞概念炒作。" 此外，神州泰岳持续加大创新投入，以创新促提升。据介绍，神州泰岳手游团队率先将生成式人工智能 技术融入美术等生产环节，大幅提升游戏制作效率；在软件板块，神州泰岳将软件开发与AI、5G、云 计算、大数据、物联网等数字技术紧密融合，为企业级客户提供数智化产品和解决方案。 同时，张开彦强调，维护投资者利 ...

Core Insights - Shandong Province aims to cultivate over 300 key audiovisual enterprises by 2027, establish more than 20 provincial audiovisual industry bases, and 2 national bases, with an annual revenue target of 200 billion yuan for the audiovisual industry [1][2] Group 1: Industry Development - The audiovisual industry in Shandong is defined as a system centered on audiovisual content, encompassing production, distribution, related equipment, technology, and services, and is a significant part of the digital economy and cultural industry [1] - The total revenue of Shandong's audiovisual industry is projected to reach approximately 130 billion yuan in 2024, reflecting a year-on-year growth of about 11.6% [1] Group 2: Technological Innovation - Shandong is accelerating the application of generative artificial intelligence, virtual digital humans, and intelligent audio-video production technologies, with high-tech continuously empowering audiovisual product creation [2] - Recent productions such as the TV series "South and North" and the documentary "The Great River" have all met 4K ultra-high-definition standards, addressing public demand for high-quality audiovisual products [2] Group 3: Economic Performance - In the first quarter of this year, the added value of Shandong's electronic information manufacturing industry grew by 28.6%, surpassing the national average growth of 17.1% [3] - The Shandong Provincial Broadcasting Bureau plans to accelerate the research and application of new technologies to extend business chains and expand the industrial ecosystem [3]

Core Insights - The rapid development of generative artificial intelligence technology has made synthetic data an essential component for training large models like GPT series. However, uncontrolled use of synthetic data can lead to "model collapse," significantly degrading model performance and generalization to real-world data [1][2][6]. Group 1: Challenges of Synthetic Data - The phenomenon of "Non-iterative Collapse" occurs when a high proportion of synthetic data is mixed into training data, leading to a significant drop in model performance even after a single pre-training session [6]. - Synthetic data has two structural defects compared to human-generated data: a lack of low-frequency and long-tail samples, which hinders the representation of language diversity, and an over-concentration of language features, increasing the risk of model overfitting [13]. Group 2: Token-Level Editing Method - The Token-Level Editing method introduces fine-grained "micro-editing" operations on real data instead of generating entire segments, creating more stable and generalizable "semi-synthetic" data, thus mitigating the risk of model collapse [3][10]. - The editing process retains the long-tail structure of the original data while only adjusting "overconfident" tokens, ensuring that the model maintains coverage of the real data distribution and avoids feature over-concentration [11][15]. Group 3: Theoretical Results - The testing error of the Token-Level Editing process has a finite upper bound, preventing model collapse, and the error does not increase with the number of iterations [12][16]. - The theoretical framework indicates that even in multi-round training, Token-Level Editing can mathematically prevent unbounded error growth, establishing a "theoretically non-collapsing" data augmentation path [16]. Group 4: Experimental Validation - The effectiveness of Token-Level Editing was validated through systematic experiments across three key stages of language model training: pre-training, continual pre-training, and supervised fine-tuning [17]. - In the pre-training phase, models using edited data outperformed those using purely synthetic data, with an average task score increase of +0.36 percentage points on benchmarks like PIQA, BoolQ, and Winogrande [18]. - In the continual pre-training phase, significant cross-domain generalization improvements were observed, such as a +13.6% accuracy increase in the PubMedQA task [18]. - During the supervised fine-tuning phase, the method demonstrated strong robustness in complex tasks, with LLaMA-3 showing an average improvement of +0.4 to +0.5% [18].

本报讯 （记者金婉霞）为贯彻落实中国证监会《上市公司监管指引第10号——市值管理》、上海市国资委《关于加强我市 国有控股上市公司市值管理工作的若干意见》等文件精神，推动上海市国有控股上市公司在做强做优做大、提升内在价值的同 时，进一步强化投资者关系管理、提高信息透明度、增进市场认同，上海市国资委联合上交所，于2025年5月9日下午共同举办 上海国有控股上市公司2024年度集体业绩说明会。 本次会议汇聚了包括上海国际港务（集团）股份有限公司、上海华谊集团股份有限公司、 关于股东回报规划及落地保障，财务总监李静强调公司十分重视股东回报，将严格落实相关规则要求，制定利润分配政 策，积极实施利润分配。未来，公司将持续依据相关规则，做优做强主业，推动积极、稳健的股东回报政策。 数据港发言环节，总裁王信菁回顾了数据港十五年的发展历程，表示公司在上海市区两级国资委、上交所及广大投资者的 支持与呵护下，逐渐从一家初创企业成长为行业领先的第三方数据中心服务商。"保持战略定力、胸怀创业梦想、坚持实干笃 行"是数据港持续发展的关键要素，王信菁指出，未来数据港将继续着眼国家数字经济战略和国企改革顶层布局，勇担"双百企 业"发展重任， ...

上海市新增5款已完成登记的生成式人工智能服务 智通财经5月13日电，上海市新增5款已完成登记的生成式人工智能服务，累计已完成85款生成式人工智 能服务登记。 ...

Group 1 - The core viewpoint of the news is that the Shanghai State-owned Assets Supervision and Administration Commission and the Shanghai Stock Exchange held a performance briefing for state-controlled listed companies, focusing on their operational performance and future strategic planning [1] - DataPort's President emphasized the importance of maintaining strategic determination and practical execution for sustainable development, aligning with national digital economy strategies and state-owned enterprise reforms [1] - DataPort's management highlighted significant improvements in operational performance, management quality, and operational efficiency for 2024, driven by the implementation of intelligent management platforms [1] Group 2 - In response to being recognized as a "Double Hundred Enterprise," DataPort plans to propose multiple reform measures focusing on strategic customer foundations and innovative development solutions [2] - DataPort's financial director stated that the company prioritizes shareholder returns and will implement a profit distribution policy in accordance with relevant regulations [2] - The company aims to deepen its integration into the national digital economy strategy, capitalizing on the growing demand for data centers driven by the rise of generative artificial intelligence [2]

Core Insights - The "2025 China Listed Companies Brand Value List" was successfully held in Shanghai, where Shenzhou Digital was awarded the "2025 Annual Private Leading Brand" for its brand influence in the digitalization field [1] Group 1: Event Overview - The event was organized by the Daily Economic News and supported by Tsinghua University's Economic Management School, focusing on the theme "Digital Intelligence Upgrade, Brand Resonance" to explore brand evolution paths driven by technology [3] - The importance of brand building as a key element for companies to establish core competitiveness and seek long-term development was emphasized in the context of a rapidly changing global business landscape [3] Group 2: Company Strategy and Innovations - Since its establishment in 2000, Shenzhou Digital has adhered to the mission of "Digital China," focusing on leading ideas, technology, and practices to promote industrial digital transformation [3] - The company has developed a full-stack AI capability under the guidance of its AI-driven cloud integration strategy, integrating AI technology with cloud-native and digital-native approaches [3] Group 3: Product Offerings and Market Response - Shenzhou Digital's KunTai series of AI intelligent computing servers, based on Kunpeng and Ascend processors, are designed for deep learning, model development, and AI training services, featuring high computing density and energy efficiency [4] - The company launched the KunTai Cube "all-in-one toolbox" and the DeepSeek version of the KunTai integrated machine to support private deployment and provide cost-effective intelligent upgrade solutions [4] - Shenzhou Digital secured an 800 million yuan order for server procurement from China Telecom for the 2024-2025 period, contributing to the construction of a domestic computing foundation for operators [4] Group 4: Future Directions - Shenzhou Digital aims to continue deepening its focus on digital transformation, leveraging its full-stack AI capabilities to create greater value for clients and help more enterprises build core competitiveness for the future [5]

最终，开福区法院判决网络公司立即停止传播某电视剧内容，赔偿原告经济损失及合理支出80万 元，二审维持原判。 来源：人民法院报 对此，被告某网络公司辩称，其提供的是剪辑工具，是技术服务提供者，用户对素材有删除、排序 的绝对控制权。被告某网络公司仅是根据用户输入的内容进行全网搜索后提供给用户，因而，用户在使 用该软件进行视频剪辑时，被告某网络公司仅提供技术服务，系中立的工具属性，不构成侵权。 法院审理认为，当前生成式人工智能技术发展迅速，但尚无专门法律条文明确规范其服务提供者的 权利义务。因此，为合理界定生成式人工智能服务提供者的法律地位，仍应依据《中华人民共和国民法 典》第一千一百六十五条之规定，以"过错责任"为基础界定责任。若仅因用户生成侵权内容就认定服务 提供者侵权，则可能对其施加过重义务。为平衡知识产权人与服务提供者的利益，需重点判断服务提供 者在内容生成阶段是否存在过错，综合考虑技术发展阶段、行业共识及技术可行性，判断其是否已建立 合理的知识产权侵权风险防范体系。 法院认为，首先，本案中被告公司没有获得权利人的授权，就通过某软件的AI功能，向用户提供 了3秒至7秒的该剧片段，涉案作品素材片段属于视听作品 ...

原标题：防范学生过度依赖生成式人工智能 《使用指南》明确了五大应用原则。一是坚持育人导向，强化素养本位。引导学生在生成式人工智 能技术使用过程中形成适应智能时代发展的价值观、必备品格和关键能力。二是坚持教育公平，尊重学 生差异。确保技术工具平等地惠及不同地区和个体，无障碍地服务有身体或认知障碍的特殊需求群体。 三是坚持价值引领，确保技术向善。营造健康有序、向上向善的人工智能教育生态。四是坚持需求驱 动，持续稳步推进。结合学校实际，有目的、有计划、有组织地选择、匹配与应用生成式人工智能技术 工具。五是坚持底线思维，保障安全可控。构建覆盖数据安全、伦理审查、内容监管和风险防控的全链 条保障机制，筑牢生成式人工智能技术在教育领域应用的安全底线。 为防范学生因过度依赖生成式人工智能而弱化独立思考能力，《使用指南》从制度规范、教学引导 与角色定位等维度建立系统性防范机制。一方面，明确禁止学生直接复制人工智能生成内容作为作业或 考试答案，并限制在创造性任务中滥用人工智能，从源头上杜绝"代劳式"使用行为。另一方面，强化教 师的引导职责，要求教师在教学实践中积极开展批判性思维训练，培养学生对技术输出内容的质疑精神 与甄别能力 ...

南都讯记者吕虹发自北京12日，南都记者从教育部获悉，《中小学人工智能通识教育指南(2025年版)》 和《中小学生成式人工智能使用指南(2025年版)》(以下简称《使用指南》)正式发布。其中，《使用指 南》明确，禁止学生直接复制人工智能生成内容作为作业或考试答案，并禁止小学阶段学生独自使用开 放式内容生成功能。同时该指南强调，教师不得将生成式人工智能作为替代性教学主体。 据教育部相关负责人介绍，此次施行的两项指南是科学规范推进人工智能全学段教育、进一步落实培养 具有人工智能素养的创新人才的重要举措。《使用指南》重点围绕生成式人工智能在中小学教育中的应 用场景，明确各学段使用规范，确保技术安全、合理、以有效地辅助教学、促进学生个性化学习、推动 教育管理智能化。而《中小学人工智能通识教育指南(2025年版)》则旨在构建一套科学完备的人工智能 通识教育体系，通过螺旋式课程设计以及"课程重构、资源融通、评价创新、师资赋能"的联动机制，推 动人工智能教育从局部试点转向全域覆盖。 南都记者了解到，《使用指南》明确了AIGC在中小学教育中的三大类核心应用场景，即促进学生成 长、辅助教师教学和支撑教育管理。 针对《使用指南》提 ...