Workflow
预训练模型
icon
Search documents
色情视频被恶意“炮制” 几元就能买到预训练模型警惕AI换脸技术滥用滋生黑灰产
Xin Lang Cai Jing· 2026-01-07 21:21
Core Viewpoint - The rise of virtual synthesis technology, particularly AI face-swapping and deepfake, has led to the creation of unethical pornographic videos, posing serious threats to personal safety and societal morals, necessitating stronger governance [1]. Group 1: Incidents and Impact - A case study of a female streamer, Xiao Yu, illustrates the dangers of AI face-swapping, where her face was maliciously placed on pornographic videos, leading to public backlash and personal distress [2]. - Xiao Yu's experience is not isolated; many women have been victimized by similar practices, with illegal groups offering services to create such videos for profit [3]. Group 2: Technology and Accessibility - Deepfake technology involves using AI to generate false content by combining personal attributes like voice and facial expressions, with AI face-swapping being the most common application [4]. - The process of creating high-quality deepfake videos requires minimal resources, including just a few photos of the victim and access to pre-trained models, which are readily available on various online platforms [5][7]. - The ease of access to pre-trained models for deepfake creation highlights significant vulnerabilities in the current online environment [8]. Group 3: Law Enforcement Challenges - Law enforcement faces unprecedented challenges in combating AI-generated content, particularly due to the anonymity and technical sophistication of offenders, making evidence collection difficult [9]. - Traditional methods of evidence gathering are ineffective against AI crimes, necessitating new strategies and collaboration with international law enforcement [10].
上海:打造人工智能国际开源社区 推动开源平台持续丰富预训练模型、人工智能应用等多样化资源
Core Viewpoint - The Shanghai Municipal Government has issued an implementation plan to strengthen the open-source ecosystem, focusing on the development of an international open-source community for artificial intelligence [1] Group 1: Open-source Community Development - The plan aims to create an international open-source community for artificial intelligence [1] - It emphasizes the continuous enrichment of open-source platforms with diverse resources such as pre-trained models, training datasets, development tools, and AI applications [1] - The initiative seeks to enhance the operational services and commercial promotion capabilities of open-source projects to meet the full chain demand of model development, training, testing, hosting, and operation [1] Group 2: Support and Resources - The plan supports the release of overseas sites for open-source platforms and the development of multilingual modules [1] - It actively organizes overseas activities to promote the open-source community [1] - The initiative includes strengthening support through computing vouchers and model vouchers, as well as providing policy-based computing resources in an orderly manner [1]