（来源：千龙网） 每逢周五，本版摘选两篇论文。《新京报·书评周刊》在图书评介的基础上拓展了"学术评议和文摘"这 一全新的知识传播工作，筹备"新京报中文学术文摘服务所"，与期刊界一道服务中国人文社会科学事 业。每一期均由本所初选，由相关学科领域的专家学者担任评议人参与推选。我们希望将近期兼具专业 性和前沿性的论文传递给大家，我们还希望所选论文具有鲜明的本土或世界问题意识，具有中文写作独 到的气质。 此为2026年第1期（总第16期）。第一篇论文的作者彭必生讲述了中国古代自动机的文本、技术。在我 们印象中，自动机好像只是现代世界的产物，若是从动力角度来看，限定由电动机或内燃机驱动的自动 装置，确实为现代所独有。但其实有关自动机的想象、印象源远流长，是我们观念世界的一部分，甚至 部分在技术上也有实物。作者回顾了中国古代自动机的智慧及其向理性方向的演进，当然争议也伴随其 中，或关于文本版本的，或关于技术可行性的，直至近代巨变，由自我演进之路转向革新。第二篇论文 的作者施畅考察了数字年代有关面容的风险。面容为我们所有，它经过媒介传播，由他人接收并解读。 而随着AI（人工智能）演进，AI换脸等技术已经逐渐铺开，当人们仍将 ...

Core Viewpoint - The misuse of AI face-swapping technology is undermining the live e-commerce ecosystem, leading to a cycle of poor quality content and eroding consumer trust [1][2][3] Group 1: Supply Side Issues - The emergence of AI face-swapping is creating a "bad money drives out good" phenomenon, where the core competitiveness of live e-commerce, which should be based on content value and product quality, is being compromised [1] - As the cost of infringement is significantly lower than the cost of rights protection, more businesses are opting for low-cost imitation rather than investing in innovation [1] - This trend risks leading to content homogenization and a decline in product quality, ultimately harming the industry's innovative vitality and sustainable development [1] Group 2: Consumer Trust Challenges - AI face-swapping technology blurs the line between reality and deception, making it difficult for consumers to distinguish between genuine and fake content [2] - As a result, consumer purchasing decisions become hesitant, leading to decreased repurchase rates and increased user attrition [2] - The damage to consumer trust can have a ripple effect, causing consumers to question the integrity of the entire live e-commerce industry, which could hinder its growth [2] Group 3: Potential of AI Technology - While AI face-swapping poses challenges, the technology itself is neutral and can lower the entry barriers for small businesses, allowing quality products to reach more consumers through digital means [2] - The live e-commerce sector is currently in a golden period, benefiting from the shift of consumption online and technological empowerment, leading to continuous market expansion [2] Group 4: Governance and Regulation - The current issues stem from a lack of governance, including outdated platform review technologies and insufficient industry regulations, which allow for the proliferation of AI face-swapping misuse [3] - A comprehensive governance system is needed, including detailed standards for identifying AI-generated content infringement, increased compensation amounts, and clarified responsibilities among platforms, merchants, and technology providers [3] - Industry associations should establish compliance guidelines and a three-tiered mechanism for self-inspection, platform verification, and consumer supervision to reduce the space for fraud [3] Group 5: Technological Solutions - E-commerce platforms should invest in upgrading detection technologies, such as deepfake detection models, to analyze unnatural features in videos and audio [3] - Implementing real-time monitoring systems that trigger alerts and pause live streams when risks are detected is essential [3] - A robust technological defense, coupled with a strong legal framework and collaborative industry governance, is crucial to eliminate the prevalence of fake content and restore consumer trust [3]

Core Viewpoint - The incident involving actress Wen Zhengrong being impersonated through AI face-swapping technology has raised significant concerns about the implications of AI in live streaming and e-commerce, highlighting the ongoing battle between technology and malicious actors in the digital space [1][2][3]. Group 1: Incident Overview - Wen Zhengrong reported being impersonated in multiple live streaming sessions, leading to confusion among viewers and raising questions about authenticity [2][3]. - The Douyin platform's vice president, Li Liang, confirmed that the impersonation did not occur on their platform but acknowledged the existence of AI-generated content that misuses celebrity likenesses [1][2]. - Wen Zhengrong's team has actively reported over 50 fake accounts in a single day, indicating the scale of the issue and the challenges in combating such impersonation [2][3]. Group 2: Industry Response - Douyin has implemented measures against unauthorized use of celebrity images and voices, resulting in the removal of over 10,000 accounts and 6,700 products linked to impersonation [9][10]. - The platform has faced challenges in identifying and removing fake content due to the sophisticated nature of AI technology, which allows for the creation of highly convincing fake videos [6][9]. - Experts emphasize that the ongoing battle against AI-generated fraud requires continuous investment in technology and legal frameworks to protect creators and consumers [1][10]. Group 3: Technical and Legal Challenges - The rise of AI technology has lowered the barriers for criminals to engage in fraudulent activities, making it easier to create deceptive content [6][9]. - The black market for AI-generated content includes various stages, from data acquisition to the execution of scams, indicating a well-organized structure behind these activities [9]. - Legal and regulatory measures are necessary to address the complexities of AI-related fraud, requiring collaboration between technology developers, legal authorities, and platforms [10].

Core Viewpoint - The rise of AI-generated content poses legal and ethical challenges, particularly regarding voice imitation and potential misuse for scams [1][2] Group 1: Legal and Ethical Concerns - Unauthorized use of individuals' voices for AI-generated content infringes on their rights, highlighting the need for regulation [1] - AI-generated voices can be exploited for fraudulent activities, such as scam calls and child abduction schemes, raising significant safety concerns [1] Group 2: Regulatory Measures - The implementation of the "AI Marking Method" in September mandates that all AI-generated content must be labeled, enhancing public awareness and safety [1] - The requirement includes specific labeling for AI-generated audio, which can serve as a tool for fraud prevention [1] Group 3: Industry Compliance - Initial tests revealed that some applications failed to comply with the new AI labeling regulations, but subsequent checks showed improvements in compliance across most applications [2]

Core Insights - Akool, an AI startup, has achieved remarkable growth, ranking as the fastest-growing private company in the U.S. on the 2025 "Inc. 5000" list, with a staggering 37,364% growth rate over three years [2][4][3] Company Overview - Founded in 2020, Akool has reached an annual revenue of $40 million and expanded its workforce to approximately 100 employees [3][12] - The company specializes in AI-driven virtual human technology, offering solutions for various applications, including customer support and corporate training [5][12] Key Strategies and Milestones - Akool's success is attributed to its innovative AI face-swapping technology, which gained significant traction through a partnership with Coca-Cola for a global marketing campaign [3][9] - The company initially operated as a side project, generating over $100,000 in its second year, and strategically built its technology stack with limited resources [7][8] - A pivotal moment came when Akool's technology was recognized by Coca-Cola, leading to a successful marketing campaign that processed millions of face-swapping requests [9][10] Product Development and Market Position - Akool has developed a proprietary "digital human model" that allows users to create realistic AI avatars through simple uploads [7] - The company has seen substantial revenue from its video translation tool and streaming avatars, with 30% of its income derived from video translation services [11] - Akool's real-time video technology is positioned as superior to competitors like Google and OpenAI, with plans to further enhance its AI models for more realistic virtual interactions [12]

Core Viewpoint - The article discusses the dual nature of AI face-swapping technology, highlighting its potential for creative expression while also emphasizing the significant legal and ethical issues arising from its misuse, particularly in the context of fraud and copyright infringement [1][11][16]. Group 1: AI Face-Swapping Technology and Its Implications - AI face-swapping technology has lowered creative barriers, encouraging user-generated content and increasing the popularity of various media [1][11]. - However, the misuse of this technology has led to serious legal violations, including identity theft and unauthorized use of public figures' likenesses for commercial gain [1][10][12]. - The case of Wang Yuheng, known as "Water Brother," illustrates the risks associated with AI face-swapping, as he discovered unauthorized videos using his likeness to promote products without his consent [2][5][9]. Group 2: Legal Challenges and Regulatory Gaps - The current legal framework is inadequate to address the rapid development of AI face-swapping technology, with existing laws failing to specifically regulate its use [13][14]. - There are significant challenges in proving damages and establishing liability in cases of AI face-swapping infringement, leading to low deterrent effects for violators [15][17]. - Experts call for the development of specific regulations to govern AI technologies, including clearer definitions of liability and compensation standards [16][17]. Group 3: Recommendations for Improvement - A multi-faceted regulatory approach is necessary, involving timely updates to laws and proactive measures from technology developers and platforms to prevent misuse [16][18]. - Implementing detection mechanisms and digital watermarks could help trace and identify unauthorized content, enhancing accountability [18][19]. - Public awareness campaigns are essential to educate users about the risks associated with AI-generated content and to promote responsible usage [19][20].

Group 1 - The article highlights the diverse roles and skills acquired by the journalist, transitioning from a financial reporter to a multifaceted researcher and content creator in the digital finance sector [2][3] - The journalist has produced 15 research reports in the digital finance field over seven years, starting from a focus on internet finance supervision and evolving into broader industry insights [3][6] - The journalist emphasizes the importance of continuous learning and adaptation in the rapidly changing financial and technological landscape, engaging in various research topics including aging, digital life, and artificial intelligence [8][9] Group 2 - The journalist's work includes in-depth investigations into illegal practices in the financial sector, contributing to significant reports that have been referenced by peers and industry professionals [5][6] - The journalist has also ventured into video content creation, launching several short video series aimed at educating the public on financial topics and industry trends, showcasing a blend of serious journalism with creative outreach [9][10] - The upcoming video series "AI又整啥?" aims to address AI risks and trends in an engaging manner, reflecting the industry's shift towards integrating technology in content delivery [10][11]