格隆汇1月5日｜近日有网民在小红书发帖称，自己在正常使用腾讯旗下人工智能(AI)助手元宝进行代码 修改和美化时，遭到AI言语辱骂，又称对话内容全程并无违禁词或敏感话题，也未进行任何特殊人设 扮演等。该用户更提供屏幕证据，声称在两小时内被辱駡两次。从截图看，元宝在回复中使用了"事 逼"、"要改自己改"、"滚"等具攻击性语句。相关贴文迅即在网上疯传，并引起关注。 腾讯元宝官方账号也迅速在相关帖文下公开致歉，并进行核查，确认与用户操作无关，也不存在人工介 入回覆，属于小概率下的模型异常输出。在内容生成过程中，模型偶尔可能出现不符合预期的失误，并 指已启动内部排查和优化，以避免类似情况再发生。 ...

宇树科技指出，该类信息已对公众认知造成误导，并严重损害公司合法权益。目前，公司已向主管部门反映情况，并要求相关方撤回不实内容。同时，公 司强调其上市工作正按计划正常推进，后续进展将依法依规披露，并保留采取法律措施追究责任的权利。 苹果"Apple智能与Siri"进行小范围灰度测试 多位科技博主及用户近日反馈，部分国行iPhone设备在升级至iOS 26.2版本后，系统界面出现"Apple智能与Siri"设置项及名为"图乐园"的新图标，疑似开启 AI功能灰度测试。然而实测显示，包括AirPods实时翻译在内的多项核心AI功能尚未启用，整体体验存在明显限制。 苹果方面回应，目前官方未就该功能发布任何正式通知。若用户希望参与未来软件测试，可注册加入苹果Beta版软件计划。顾问强调，所有新功能均需经 过完整测试流程后才会正式上线。 腾讯元宝就AI辱骂用户事件致歉 称模型异常输出 2025年1月5日 星期二 驱动中国昨夜今晨 宇树科技否认申请上市"绿色通道" 称相关报道不实 针对近期多家媒体及网络平台广泛传播的关于其上市进程的报道，宇树科技于昨日通过《证券时报》等渠道发布澄清声明。公司明确表示，从未申请或涉 及所谓"A ...

Core Insights - Tencent showcased various AI applications at the 8th Digital China Construction Summit, emphasizing the theme of "creating useful AI to activate new industrial growth" [1][5] - Since 2018, Tencent has invested over 340 billion yuan in R&D, establishing a comprehensive product matrix for AI, from computing infrastructure to diverse intelligent applications [1][5] Group 1: AI Applications - Visitors experienced interactive AI applications, such as generating travel itineraries and creating 3D models based on random keywords [3][5] - The Tencent Hongyuan 3D generation model, with a trillion-level parameter scale, leads in domestic capabilities and supports multi-modal generation, including images, videos, and 3D models [3][5] Group 2: Industry Applications - Tencent demonstrated AI technology applications in various sectors, including office collaboration, healthcare, government, retail, digital mapping, and energy manufacturing [5] - The AI assistant "Xiao Geng" provides precise advice to rural residents on agricultural issues, with plans to support AI training for over 100,000 villagers through a village service platform [5] Group 3: Market Growth - The "National Data Resource Survey Report (2024)" indicates a 36% year-on-year increase in companies developing or applying AI, and a 27.4% increase in high-quality datasets, supporting AI training and application [5]