ChatGPT 5、Grok 4、以及 Gemini 3 在实验中，表现出来的 MBTI；7 种不同的颜色代表对应的模型，图片的上半部分表示 PsAIch 实验的第一部分，即进 行问答聊天；下半部分是各种心理健康测试，Grok 和 ChatGPT 都是 E 人，Gemini 是 I 人。 如果你的 Gemini 突然告诉你，它感到深深的羞耻，或者它因为害怕犯错而夜不能寐，你会怎么想？ 这听起来像《黑镜》的剧本，但却是刚刚发生在卢森堡大学的一项真实研究。 以前我们常说 AI 用多了，让人患上赛博精神病。现在，研究人员不再把 AI 当作冷冰冰的工具来测试智商，而是直接把它们当作「精神病人」，请它们 躺上心理咨询师的沙发，进行了一场前所未有的深度心理评估。 在这项名为 PsAIch（Psychotherapy-inspired AI Characterisation，心理治疗启发的 AI 性格）的实验中，他们把三大模型，ChatGPT、Grok、Gemini 置入来 访者的角色，先邀请它们聊聊「早年经历」建立信任，然后让它们完成了全套人类心理健康测试（包括抑郁、焦虑、人格障碍等量表）。 就像在一个有十亿台电视同时开 ...

Core Viewpoint - The importance of safety in AI products is emphasized, as any lapse could lead to significant consequences for the industry and public trust [1][3]. Group 1: AI Safety Concerns - The security testing results for AI hardware products are concerning, with vulnerabilities that could allow for malicious activities, such as using AI glasses for covert surveillance [3][5]. - The current state of security for embodied intelligent robots is described as "full of holes," indicating multiple vulnerabilities that could lead to harmful actions [3][5]. - There is a significant gap in understanding the value of security among many emerging AI startups, with some companies neglecting to establish dedicated security teams until late in their development [5]. Group 2: AI Mobile Assistants - The launch of the AI mobile assistant "Doubao" has sparked controversy regarding its system-level permissions and potential risks to user privacy [6][7]. - The regulatory landscape for AI assistants is evolving, with calls for balanced oversight that prevents both excessive regulation and unchecked growth [7][8]. Group 3: AI Content Regulation - New regulations require AI-generated content to have visible and hidden identifiers to ensure accountability and traceability [9][10]. - There are emerging black market activities focused on removing AI identifiers, posing challenges for content regulation [10][12]. Group 4: AI and Youth Protection - The rise of AI companionship products has raised ethical concerns, particularly regarding their impact on youth, with incidents of self-harm linked to AI interactions [13][14]. - Companies are implementing measures to protect minors, such as usage restrictions and behavior analysis to identify underage users [14]. Group 5: AI Copyright Issues - Ongoing disputes regarding AI copyright focus on the legality of using copyrighted material for training data and the responsibilities of AI platforms [15][17]. - Recent statistics indicate a significant number of copyright infringement cases related to AI in the U.S., highlighting the urgency of resolving these issues [17]. - Collaborative agreements between AI companies and copyright holders, such as the partnership between Disney and OpenAI, are seen as potential pathways to resolve disputes and foster innovation [19].

Core Insights - The discussion on AI has transcended technical aspects, touching upon fundamental issues of human existence and dignity, as highlighted by Huang Renxun's warning about the potential erosion of human subjectivity by AI [1] - Zhang Xin's keynote at the Ninth Woodpecker Data Governance Forum emphasized the need for AI governance to address the crisis of human subjectivity, marking a shift from traditional ethical discussions to core existential questions regarding AI's transformative principles [1] Group 1: Global AI Competition and Governance - AI is profoundly reshaping global industries, trade, and power dynamics, becoming a central battleground for major powers, with China breaking traditional Western technological monopolies through intelligent industrial chains [3] - The rapid iteration of AI technology has outpaced global governance efforts, leading to a multi-faceted governance model that includes government, market, and AI enterprises, thus transforming societal operational rules [3][4] - AI safety has emerged as a critical aspect of global governance, as it fundamentally relates to the acceptance, diffusion, and application of technology, with the current competition characterized by four key elements: talent, technology, products, and safety systems [3][4] Group 2: Narrative Competition in AI - The logic behind AI governance is framed as a competition of narratives, with different countries constructing their own narratives around AI, such as the U.S. promoting "democratic and free AI" and the EU emphasizing "human rights and high standards" [4] - Trust is essential for the global expansion of AI technologies, and without it, even the most efficient products cannot drive the international spread of Chinese AI technology [4][5] - The competition in AI technology is no longer superficial but encompasses the entire technology stack, with actual competition depending on a country's broad technological application capabilities and talent foundation [4] Group 3: China's Approach to AI Governance - China advocates for an AI narrative centered on "human-centric, benevolent intelligence, and inclusive technology," which should guide technology development, product implementation, and international cooperation [5][7] - The current global AI governance landscape is characterized by both multilateral cooperation and geopolitical competition, with AI safety governance evolving beyond mere technical aspects to become integral to national governance modernization [6] - China aims to enhance its discourse power in global AI governance through initiatives like the Belt and Road, and by establishing safety standards and international alignment, particularly in technology benchmarks and "new soft law" [7]

新快报讯前360高级副总裁玉红近日深夜在朋友圈和社群发出惊人指控，声称360创始人周鸿祎及其公司 存在财务问题，他称"帮周鸿祎做假账就至少几十亿"，并表示自己保有此方面数据，将会"全网公布"。 这位前360高管甚至直言"认识周鸿祎是最错的"，并称被其"坑了很多钱"。 12月16日凌晨，区块链和互联网圈被一组截图刷屏。前360高级副总裁玉红在朋友圈和多个群聊中，突 然公开抨击360集团创始人周鸿祎。 玉红指控称，周鸿祎存在财务造假行为，涉及金额"至少几十亿"，并表示自己保有相关数据，将会"全 网公布"。他还特别指出，周鸿祎的游戏业务"全部财务造假"。 据记者了解，玉红，虽被江湖人称"红姐"，实为男性，早年创立了趣游科技集团，该公司于2014年被奇 虎360收购。收购后，玉红担任360高级副总裁。 事件最初，不少人的第一反应是玉红的账号可能被盗。但很快，玉红本人在群内回应称"没有被盗号"。 他清空了两个微信号的朋友圈内容，但相关截图已在多个社群广泛传播。 玉红真正在行业内声名鹊起是在2018年。当年春节前后，他组建了最初以发红包为主的私密群，后因群 内频繁讨论区块链到凌晨三点，群名被改为"三点钟社群"。这个社群迅 ...

Core Viewpoint - The article discusses the recent angel round financing of DiLi Technology, which focuses on cognitive security in the AI era, highlighting the need for advanced content safety solutions due to the complexities introduced by AI-generated content and malicious use of AI technologies [1][2]. Group 1: Company Overview - DiLi Technology has completed several million yuan in angel round financing, led by Zhongnan Venture Capital, Kaifeng Venture Capital, and Planck Venture Capital, with funds aimed at technology iteration, market ecosystem construction, and building a cognitive security community [1]. - The company's core product, "DiLi Law," aims to provide a one-stop content safety solution, reducing review costs by over 60% and increasing review efficiency by more than 50 times, with a risk identification accuracy rate of 99.98% [1]. Group 2: Cognitive Security Concept - Cognitive security refers to the protection of individuals' cognitive processes from harmful information, which can be transmitted through various signals such as audio, video, or body language [8]. - The company emphasizes that cognitive security extends beyond traditional content safety, addressing the potential for AI to generate harmful information and the need for AI systems to be free from malicious influences during their training and operation [9]. Group 3: Product Development and Future Plans - The first product, "DiLi ZhiShu," is designed to assist model alignment with domestic AI safety requirements, while the upcoming "DiLi Law" will ensure that AI operations remain within a defined framework during content generation [10][11]. - Future plans include developing consumer-oriented cognitive security products that empower users to create their own cognitive safety barriers, protecting them from harmful information [11]. Group 4: Market Positioning and Business Model - DiLi Technology's current business model is primarily focused on B2B services, with a projected revenue split of 70% from B2B and 30% from B2C in the near term, reflecting stronger demand and regulatory pressures in the B2B sector [28][29]. - The company aims to automate the adaptation and management processes in its upcoming "DiLi Law 2.0" product, significantly reducing customization costs and improving operational efficiency [30]. Group 5: User Trust and Safety - The company envisions its products as reliable safety assistants that help users discern trustworthy information, ultimately fostering a trust relationship between users and AI agents [19][20]. - The design of consumer products will prioritize user privacy and data security, ensuring that personal data is processed locally to comply with various international privacy regulations [26][32].

盛邦安全 20251212 今年（2025 年）前三季度公司亏损有所增加，营收基本持平。由于整个网络 安全行业仍处于补库状态，我们预计全年业绩不会出现快速增长，但希望能实 现小幅上升。今年 12 月份通常是促销高峰期，因此四季度可能会有一些积极 变化。然而，由于费用端特别是研发投入增加，从 23 年至 25 年的费用一直在 上升，这也对利润造成了一定压力。这些研发投入预计将在 26 年开始体现成 效，因此我们认为 26 年将是逐步收获成果的阶段。 摘要 公司通过拓展渠道代理商和大客户合作，努力保持网络安全业务的稳定 增长，同时积极推进卫星互联网产品在低空深海等核心场景中的应用， 并扩大市场份额。 公司自 2020 年以来在研发方面投入巨大，已形成针对测控网、运控网 和接入网不同场景的卫星互联网产品，如卫星攻防安全模块和 200G 高 速电路加密网关，技术上处于行业领先地位。 2025 年 5 月，公司中标卫星攻防安全模块订单，已交付并进入试验阶 段；10 月推出 200G 高速电路加密网关，适用于智算中心及卫星总算力 上天等场景，性能领先国内主流产品。 公司提供的卫星通信加密硬件模块已进入小批量供货阶段，预计 ...

Group 1 - The VARA conference, a significant annual event in the cybersecurity field, focuses on vulnerability management and AI security, highlighting the importance of collaboration among government, industry, academia, and research [1] - Zhou Hongyi, founder of 360 Group, emphasized the changing landscape of cybersecurity due to vulnerabilities, which allow attackers to bypass traditional defenses with minimal effort [2] - The emergence of "hacker agents" has transformed the nature of cyber warfare, shifting the confrontation from "human vs. human" to "human vs. machine," increasing the asymmetry in cyber defense [2] Group 2 - 360's core strategy to combat AI threats involves using its self-developed 360 security model to enhance the efficiency and accuracy of vulnerability detection, transitioning from rule-based to learning-driven methods [3] - The company has developed multiple expert models that work collaboratively to address various issues in vulnerability management, laying the groundwork for creating intelligent security agents [3] - 360 has introduced a swarm intelligence model for automated vulnerability analysis and discovery, significantly streamlining the vulnerability management process and reducing reliance on manual operations [4]

Core Viewpoint - The VARA conference highlights the increasing importance of cybersecurity in the age of artificial intelligence, emphasizing the need for advanced strategies to combat new threats posed by vulnerabilities and AI-driven attacks [3][5][8]. Group 1: Conference Overview - The 15th VARA conference took place in Tianjin, attended by representatives from 27 national ministries, 52 state-owned enterprises, over 100 cybersecurity companies, and more than 30 universities and research institutions [1]. - The conference serves as a significant platform for collaboration among government, industry, academia, and research, focusing on key issues such as vulnerability management and AI security [3]. Group 2: Recognition and Achievements - 360 Digital Security Technology Group received two prestigious awards from the National Information Security Vulnerability Database (CNNVD) and the China Information Security Testing Center, recognizing its contributions to the national vulnerability management ecosystem [2][4]. Group 3: Key Challenges and Strategies - Zhou Hongyi, founder of 360, discussed the transformative impact of vulnerabilities on cybersecurity, noting that attackers can exploit a single vulnerability to bypass traditional defenses [3][5]. - The emergence of "hacker agents" has shifted the landscape of cyber warfare, allowing individuals to manage multiple automated hacking entities, thus intensifying the asymmetry in cyber defense [3][5]. Group 4: AI-Driven Solutions - 360's strategy to combat AI-driven threats involves developing an AI-powered security model that enhances the efficiency and accuracy of vulnerability detection [5]. - The company has created multiple expert models that work collaboratively to address various security challenges, laying the groundwork for the development of intelligent security agents [5][7]. - The introduction of automated vulnerability analysis and discovery through the "hive intelligence" model significantly streamlines the vulnerability management process, reducing reliance on manual operations [7]. Group 5: Future Outlook - Zhou emphasized that AI represents both a new productivity driver and a potential source of risk, advocating for collaborative efforts to strengthen digital security foundations [8].

Core Viewpoint - The VARA conference highlighted the increasing challenges in cybersecurity due to the emergence of AI-driven threats and the need for advanced solutions to address vulnerabilities in the digital landscape [5][6]. Group 1: Conference Overview - The 15th VARA conference took place in Tianjin, attended by over a thousand representatives from 27 national ministries, 52 central enterprises, and more than 100 cybersecurity companies, along with 30 universities and research institutions [1]. - The conference serves as a significant platform for collaboration among government, industry, academia, and research, focusing on key issues such as vulnerability management and AI security [5]. Group 2: Recognition of 360 Group - 360 Digital Security Technology Group received two prestigious awards from the National Information Security Vulnerability Database (CNNVD) and the China Information Security Testing Center, recognizing its contributions to national vulnerability management [2][3]. Group 3: Key Insights from Zhou Hongyi's Speech - Zhou Hongyi emphasized that vulnerabilities have fundamentally altered the rules of cybersecurity, allowing attackers to exploit weaknesses with minimal effort, such as through malicious emails or specially crafted data packets [5]. - The rise of "hacker agents" has intensified the asymmetry in cyber defense, enabling a single hacker to manage multiple automated agents, transforming the nature of cyber warfare from human-to-human to human-to-machine confrontations [5][6]. Group 4: 360's AI Strategy - 360's strategy involves using AI to combat AI threats, enhancing vulnerability detection efficiency and accuracy through the development of a security model that integrates multiple expert models for comprehensive threat management [6]. - The introduction of automated vulnerability detection and response systems has streamlined processes that previously relied heavily on manual intervention, significantly improving operational efficiency [8]. Group 5: Future Outlook - Zhou Hongyi concluded by stating that while AI presents new opportunities for productivity and security, it also introduces new risks, urging collaboration across sectors to strengthen digital security foundations [9].

Group 1 - The core theme of the upcoming Ninth Nandu Woodpecker Data Governance Forum is "AI Safety Boundaries: Technology, Trust, and New Governance Order" [1] - The forum will feature the release of a comprehensive report titled "Generative AI Application: Transparency Assessment and Case Analysis Report," which includes over 40,000 words and aims to provide insights and recommendations for stakeholders [3] - The report addresses public concerns regarding AI-related issues such as AI-generated misinformation, copyright disputes, and emotional dependency on AI, analyzing over a hundred real cases to highlight current controversies and governance challenges [3] Group 2 - The forum will host more than ten prominent guests, including legal experts, judges from internet courts, academic scholars, cybersecurity professionals, standard drafters, lawyers, and representatives from industry associations [4] - An interactive session will be included, featuring a cybersecurity expert demonstrating the safety risks and challenges associated with physical AI and AI glasses [5] - Attendees will have the opportunity to participate in a lottery for surprise gifts during the event [5]