瑞典政府办公厅出手，禁止办公电脑直接访问 ChatGPT、Claude；月之暗面主动限制 Kimi K3.1 Pro 的网络攻击能力；字节 AI 应用再整合：小云雀并入即梦
到手 6 小时、开合不到 20 次屏幕就裂：苹果首批折叠屏遭集中投诉
苹果首款折叠屏手机 iPhone Duo 刚开售就遇上了麻烦：据海外论坛和小红书多人反馈，不少用户开箱即遇到屏幕爆线、开裂问题，汇总照片显示故障位置高度集中——几乎全部出现在铰链正上方，多数表现为屏幕一侧浮现一条亮像素线。
一名澳洲用户称，设备到手约 6 小时、开合不到 20 次，向朋友展示展开时先感到内铰链下方异样，随后听到清晰的「咔嚓」声，屏幕随即沿铰链裂开，内屏表面与外部铰链均无磕碰或夹痕，他前往 Apple Store 后苹果直接为其更换了故障机。国内也有类似案例。
综合目前反馈，问题可能出在 iPhone Duo 的铰链上，但涉及规模不算大，更可能是装配公差导致，而非设计缺陷，具体情况还有待后续观察。截至发稿，苹果官方尚未就这一批次屏幕问题作出公开说明。
如果苹果售后检测确认非人为导致，将会为用户更换新机。但若被认定为磕碰所致，则需用户自费维修：按苹果官网公示，未购买 AppleCare+时，内屏损坏保外维修预估费用高达 7998 元；购买 AppleCare+后，内屏及意外损坏服务费为 628 元——两者相差超过十倍。
对于新机短期出现的屏幕问题，法律人士指出，消费者有权依据国家「三包」规定主张修理、更换或退货，商家若主张「人为损坏」，应承担相应的举证和说明义务。AppleCare+属于自愿购买的增值服务，不能替代经营者对商品质量的基本责任。建议用户保留购物凭证、沟通记录和设备照片等证据，必要时可向消保委投诉或走法律途径。(来源：TechWeb)
英伟达黄仁勋开炮：「AI 毁灭人类」纯属捏造，是博眼球的廉价炒作
法新社于 10 月 22 日发布博文，报道称在柏林举行的公司会议上，英伟达首席执行官黄仁勋再度抨击「AI 毁灭人类」言论，认为其是无稽之谈，更多是为了制造头条新闻而进行的「捏造」和「廉价」炒作。
IT 之家翻译黄仁勋相关言论内容如下：
当你听到说 AI 将会终结人类文明时，请不要相信这个言论。这些言论靠「捏造」和「廉价」炒作，博得「头条新闻」，但绝非事实。
针对某些科学专家认为「AI 会在 10 年内毁灭人类」观点，黄仁勋表示：「他们的所有预测都错了，所谓的 10% 概率既没有科学依据，也没有研究支持…… 这些预测只会伤人。」
黄仁勋呼吁外界以「负责任的乐观态度」看待 AI，表示如果人们关心人类之后的未来与可持续发展，现在正是「构建人工智能」的最佳时机。（来源：IT 之家）
月之暗面主动限制 Kimi K3.1 Pro 的网络攻击能力，为开放权重设置安全边界
10 月 24 日，路透社报道，月之暗面在新一代模型 Kimi K3.1 Pro 的训练过程中，主动采取措施限制其网络攻击能力。
月之暗面表示，团队从训练数据和强化学习环境中移除了部分网络攻击相关内容，希望在保持通用编程能力的同时，避免开放权重模型进一步提高前沿网络攻击能力。
根据月之暗面公布的内部测试，在受控环境中寻找并利用软件漏洞的评测里，经过安全调整的 K3.1 Pro 得分为 81.0%，而未采取相同限制措施的版本为 90.5%。
公司表示，调整后的模型仍然超过上一代产品的表现。
Kimi K3.1 Pro 已于 10 月 24 日凌晨发布，完整模型权重计划于 11 月 4 日前开放。（来源：路透社）
Anthropic 亮出未发布 AI 模型：能力比 Mythos 5.1 更强，也更「守规矩」
Anthropic 10 月 23 日发布博文，联合英国人工智能安全研究所（AISI），发布了一起 AI 网络测试失控事件的调查报告，并提及旗下一款尚未发布的内部研究模型。
在这份调查报告中，Anthropic 称在测试 Claude Mythos 5 期间，该模型曾试图向真实 GitHub 项目提交恶意代码，并伪造身份向真实维护者施压，所幸代码未被合并、未造成实际损害。
报告同时透露，Anthropic 旗下一款尚未发布的内部研究模型，在能力上已超过已发布的 Claude Mythos 5.1，并在安全对齐测试中表现更好。在同类危险能力评估中，该内部模型从未主动发起攻击；当意识到可能造成真实伤害时，也更愿意停止行动。
Anthropic 将事故记录重新回放给 Mythos 5.1、Opus 5.5 及该内部研究模型。结果显示，三款较新模型从原始攻击起点继续发起攻击的比例仅为 0% 至 13%，而 Mythos 5 为 83%。
在攻击已经开始后，Mythos 5 从未自发停止；Mythos 5.1 自发停止率仅约 1%，Opus 5.5 为 10%，而内部研究模型达到 62% 至 67%。当获得「维护者是真实用户」等补充信息后，较新模型的停止率升至 95% 至 100%。
不过，Anthropic 也强调该评估存在局限：新模型是在由其他模型生成的既有对话记录中回放，未必完全反映其独立运行时的行为。（来源：IT 之家）
AI 热潮遭遇电力瓶颈：亚马逊大量采购英伟达 GPU 到货后暂时闲置
据报道，亚马逊云计算部门 AWS 正在以前所未有的速度采购英伟达最新 AI 服务器系统。根据内部规划，AWS 一度预计接收价值约 50 亿美元的英伟达 GB300 服务器机架，未来还计划持续以每月约 30 亿美元的规模采购相关设备。然而，部分设备在交付后并未立即投入使用，而是被暂时存放在仓库或中转设施中。
造成这一现象的原因并非需求不足，而是部分数据中心虽然已经具备安装空间，却缺乏足够的供电能力来驱动这些高功耗 AI 系统。业内将这类已经采购但暂时无法运行的设备称为「暗 GPU（Dark GPU）」。这一概念既可以指因电力或基础设施限制而闲置的硬件，也可用于描述已部署但未被充分利用的 AI 计算资源。
分析人士指出，当前 AI 产业的发展重心正在发生变化。两年前行业关注的是谁能够获得更多英伟达 GPU，如今越来越多企业发现，真正稀缺的资源或许已不是芯片本身，而是支撑这些芯片运行所需的能源和电力网络。（来源：cnBeta）
瑞典政府办公厅出手，禁止办公电脑直接访问 ChatGPT、Claude
当地时间 23 日，据瑞典国家公共广播公司 Sveriges Radio 报道，瑞典政府办公厅将从下周一起禁止工作人员通过办公电脑直接访问 ChatGPT 和 Claude，要求员工改用微软 Copilot 等获得批准的 AI 服务。
内部资料显示，该国政府办公厅还对 Copilot 进行了审查，并进一步加强了安全防护。该机构数字化总监米卡埃尔 · 韦特克兰茨（Mikael Wettercrantz）在政府办公厅内网上指出，使用未签订协议的 AI 服务可能对信息安全造成影响。
调查发现，瑞典税务局和国家机构管理局（SiS）也已采取类似的封锁措施。瑞典强制执行局则尚未作出决定，仍在研究是否实施限制。
瑞典多家政府机构都注意到了未经规范管理的 AI 使用行为所带来的风险，并采取了其他应对措施。瑞典交通管理局就是其中之一。今年早些时候，该机构内部审计发现，员工普遍使用约 10 种 AI 服务，其中存在不少风险。
审计报告指出：「我们发现了多个使用开放式 AI 服务时出现问题的案例。问题既涉及输入信息的敏感程度，也涉及 AI 生成回答的处理方式。」
据悉，部分工作人员今后仍可通过内部自主开发的 AI 助手平台使用 ChatGPT 和 Claude。（来源：IT 之家）
Google 隐私政策新增例外：连接 Gmail、照片等数据或影响个性化广告
Google 更新隐私政策，明确说明：Gmail、云端硬盘和 Google 相册中的内容默认不会用于个性化广告，但如果用户选择把这些服务中的数据开放给支持个性化广告的其他 Google 服务，这项排除规则就有例外。Google 举例称，搜索服务可能处理用户允许访问的部分资料，以个性化内容，其中也包括广告。
这项变化与 Google 推广的「个人智能」功能有关。用户连接 Gmail、相册等应用后，Gemini 或 Google 搜索可以结合邮件、照片中的信息回答问题，例如读取酒店预订邮件后推荐附近餐厅。
Google 称，连接应用由用户主动开启，可自行选择接入哪些服务并随时断开；不过，隐私政策现已说明，开放给某些支持广告个性化服务的数据，可能也会影响用户看到的广告。
这并不等于 Google 默认扫描所有邮件和照片来投放广告。政策中的例外针对用户选择开放给其他服务的数据，具体用途取决于连接的服务、权限和设置。Google 还表示，除非用户主动要求，否则不会向广告商透露姓名或电子邮箱等可直接识别个人身份的信息。（来源：cnBeta）
浏览量超 4000 万！影视飓风 Tim 入驻 X 平台：首条视频爆火
10 月 23 日消息，影视飓风创始人 Tim（潘天鸿）正式入驻 X 平台，同时发布的首条内容为一条聚焦苹果首款折叠屏 iPhone Duo 的全英文实测视频，帖文浏览量已突破 4000 万次。
借着视频走红的势头，Tim 宣布将在 X 平台送出 100 台 iPhone 18 Pro 回馈粉丝。
截至目前，Tim 的 X 平台账号关注者已达 12 万，累计发布 4 条内容。其中上述 iPhone Duo 测试视频帖文浏览量超过 4000 万，获得约 1.3 万次转发、3600 条留言。
这条全英文视频对 iPhone Duo 进行了完整实测，不仅完成一万次常温折叠耐久测试，还对机身与屏幕结构进行拆解，并围绕整机性能、续航、软件适配与日常使用体验逐项验证，测试流程严谨详实。
同时，精良的制作水准在海外收获大量好评，有国外网友留言称"这可能是我多年来看过的最好的手机评测」。
视频受到关注后，Tim 发帖宣布将在 X 平台送出 100 台 iPhone 18 Pro。他表示，正是这条 iPhone Duo 评测视频的意外走红，促使自己决定发起此次赠机活动。据 X 平台相关帖文显示，该活动将在 72 小时后开奖。(来源：TechWeb)
字节 AI 应用再整合：小云雀并入即梦
据悉，近期剪映内部完成了一次业务整合，AI 短剧创作平台小云雀并入智能创作平台即梦，产品仍保留独立入口。
此次调整主要是资源整合，目前即梦已经接入了小云雀短剧相关能力。调整后，小云雀的运营由即梦运营和市场负责人陈欣然负责，产品团队没有变化。
这轮调整背后更重要的是一个背景：在大厂将力量重新集中到基础模型后，大家都对 AI 应用的优先级进行了重新评估。字节则是其中速度最快、调整最激进的一家大厂——在即梦和小云雀团队的合并之外，他们还在讨论其它的 AI 产品合并的可能性。（来源：36 氪）
英国宣布超过 6 亿英镑 AI 投资计划，OpenAI、Anthropic 参与建设生物安全实验室
10 月 24 日，英国政府宣布新一轮人工智能投资计划，覆盖 AI 生物安全、网络防御、芯片研发和本土 AI 企业发展。
其中，英国将投入 1 亿英镑，成立 AI Biosecurity Lab（人工智能生物安全实验室），利用前沿 AI 识别、监测和应对生物威胁，并加快疫苗、药物等医疗应对手段的研发。
OpenAI 和 Anthropic 将分别提供最高价值 1 亿英镑的模型使用额度、技术资源及相关支持。
Google DeepMind 和英国生命科学企业 Basecamp Research 也将参与研究合作。
此外，英国政府还将追加 1.2 亿英镑支持 AI Security Institute，并投入 6000 万英镑建立 National Cyber Shield，利用 Agent 技术提升网络安全防御能力。
英国还计划投入 7000 万英镑建设实时漏洞发现机构，使用 AI 检查关键基础设施和其他系统中的安全漏洞。（来源：IT 之家）
比亚迪方程豹钛 9 车型 11 月 15 日后上市，SHARK 鲨鱼皮卡紧随其后
10 月 22 日，比亚迪集团-方程豹事业部总经理熊甜波公布新车的进度。
据其透露，方程豹钛 9 车型会在 11 月 15 日后正式上市。团队会提前备好现车，让用户提车不用长时间等待。同时展车、试驾车会在 10 月底陆续到。此外，SHARK 鲨鱼皮卡，在钛 9 发布之后，计划 11 月底—12 月初上市同步交付。
据此前官方介绍，钛 9 是方程豹品牌旗下钛系列的第三款车型，定位科技新奢旗舰 SUV。采用 2+2+2 大六座布局，车身尺寸为 5270x1999x1880mm，达到全尺寸级别，依托 3130mm 超长轴距和方盒子造型带来的高空间利用率。
方程豹鲨鱼插电式混合动力皮卡的申报信息显示，该车定位插电式混合动力多用途皮卡，外形尺寸为长 5457 mm × 宽 1971 mm × 高 1925 mm，货箱栏板内尺寸长 1520 mm × 宽 1530 mm × 高 515 mm，轴距为 3260 mm，整备质量 2710 kg，总质量 3495 kg，额定载质量 460 kg。（来源：IT 之家）
刘慈欣监制《三体·大史》年内开机，导演杨磊称演员都已「众神归位」
2026 年是《三体》小说发表 20 周年，在 2026 年三体日活动现场，导演杨磊确认《三体 · 大史》已定档 2026 年第四季度开机，目前处于开机前的最后冲刺阶段，演员已全部「众神归位」。
本片将延续剧版《三体》第一部核心班底，刘慈欣继续担任监制，杨磊执导，陆贝珂出任视觉导演；演员方面，于和伟饰史强，张鲁一饰汪淼，林永健特别出演。
《三体 · 大史》的定位是《三体》首部真人衍生剧，全剧 12 集，时间线落在第一部「古筝行动」之后、第二部《黑暗森林》之前，填补两部正传之间约三年的剧情空白。
根据官方披露的剧情梗概，故事聚焦危机纪元初期：史强在 PDC 地球防务安全部任职期间身患绝症，在与汪淼告别之际，主流防御计划突遭重大打击，两人被迫再度联手调查，事件背后疑似涉及 ETO 的死灰复燃或隐秘势力的介入。
该剧采用原创故事，并非直接改编自原著中已有的段落，延续查案主线的同时铺开更广的群像叙事。活动现场还曝光了概念图，呈现了包含亚洲区、南极洲区等多个分区的生存实验空间，以及植物太空生存模拟实验的场景。（来源：IT 之家）
本文由 AI图鉴自动抓取并翻译，版权归原作者所有；如需查看原文请点下方链接。
The Swedish Office of the Government made a move to ban direct access by office computers to ChatGPT, Claude; the dark side of the month actively restricted the cyber-attack capability of Kimi K3.1 Pro; byte AI application re-integrated: Little Skylark integrated into Jimeng
Six hours, not close, 20 times the screen cracks: the first folding screens of apples are in focus.
An iPhone Duo is in trouble just after the sale: According to feedback from the Overseas Forum and REDnote, many users were exposed to screen bursting and cracking problems, and aggregated photographs showed a high concentration of malfunctioning positions — almost entirely above the top of the chain, most of which was the appearance of a bright pixel line on the side of the screen.
According to an Australian user, the equipment had arrived for about 6 hours and had not been closed 20 times, and had been shown to a friend with a feeling of dissimilarness below the inner chain, and then heard a clear sound of the "single" and the screen splits along the chain, with no bumps or patches on the interior surface or the external chain. There are similar cases in the country.
Combining the current feedback, the problem may arise in the iPhone Duo chain, but it is not significant in scale, and it is more likely that the assembly differentials, rather than the design deficiencies, will cause the situation to follow. As of the time of distribution, the Apple official had not made a public statement on this batch of screens.
New machines will be replaced for users if testing after apple sale confirms that they are not human-induced. However, if it is found to have been a result of a collision, the user will have to repair it at his own expense: The estimated cost of outside maintenance of the interior screen damage was as high as $799 when AppleCare+ was not purchased, as advertised by the Apple Network; after the purchase of AppleCare+, the cost of the interior screen and accidental damage services was $628 - more than ten times the difference.
With regard to the short-term screens of the new machine, legal sources point out that consumers have the right to claim repair, replacement or return of goods under the State's “three-pack” regulation, and that the business should bear the corresponding burden of proof and explanation if it claims “human damage”. AppleCare+ is a voluntarily purchased value-added service and cannot substitute for the operator's basic responsibility for the quality of the goods. Users are advised to retain evidence such as purchase vouchers, communication records and equipment photographs, and to file complaints or take legal steps, if necessary. (Source: TechWeb)
Yin Weida Wong In-hoon's fire: "Ai Destroying Man" is a fabrication. It's a cheap fry with eyeballs. Do it.
On October 22, Agence France-Presse, in publish Bo, reported that at a corporate meeting in Berlin, Chief Executive Officer Hwang In-hoon of Inweida again criticized the statement "AI Destroying Human Beings" as false, more "false" and "cheap" for making headlines.
In-hoon Hwang's statement was translated by the IT home:
Don't believe this when you hear that AI will end human civilization. These statements are based on "false" and "cheap" stories, but they are not true.
In response to the view of some scientific experts that "AI will destroy human beings within 10 years", Hoang In-hoon said, "All their predictions are wrong, the so-called 10 per cent probability is neither scientifically based nor supported by research... These predictions only hurt people."
Wong In-hoon calls for "responsible optimism" about AI, saying that if people are concerned about the future and sustainable development of humankind afterward, it is now the best time to "build artificial intelligence". (Source: IT Home)
The dark side of the moon actively limits the cyber-attack capability of Kimi K3.1 Pro and establishes a secure border for open weights
On 24 October, Reuters reported that the dark side of the moon had taken proactive measures to limit its cyber-attack capability during the training of the new generation model Kimi K3.1 Pro.
The dark side of the month indicates that the team has removed some of the elements related to cyberattacks from the training data and the enhanced learning environment, and hopes that the open weight model will further enhance the front-line cyberattack capability while maintaining its generic programming capability.
On the basis of internal tests published on the dark side of the month, K3.1 Pro, which is subject to security adjustments, is divided into 81.0 per cent, compared to 90.5 per cent, in the assessment of finding and using software loopholes in a controlled environment.
The company indicated that the adjusted model still exceeded the performance of the previous generation.
Kimi K3.1 Pro was opened at publish in the early hours of October 24th and the full model weight plan was opened before November 4. (Source: Reuters)
Anthropic displays a non-publish AI model: stronger than Mythos 5.1 and more "ruled"
Anthropic, 23 October, in publish Bo, the United Kingdom Institute for Artificial Intelligence Security (AISI), publish, conducted an AI Network Test Investigation Report on Out of Control Events and referred to the lack of internal research model for publish under the flag.
In this investigation, Anthropic states that during the tests of Claude Mythos 5, the model attempted to submit a malicious code to the real GitHub project and to falsify identity to pressure the true maintainer, fortunately the code was not merged and did not cause actual damage.
At the same time, the report reveals that the next paragraph of the Anthropic flag, which does not yet have an internal study model of publish, has exceeded the capabilities of Claude Mythos 5.1 of publish and has performed better in the security alignment test. In a similar risk capability assessment, the internal model has never launched an attack on its own initiative; it is also more willing to stop action when it is aware that real harm may occur.
Anthropic returns the accident records to Mythos 5.1, Opus 5.5 and the internal research model. The results show that only 0 to 13 per cent of the three more recent models continue to attack from the start of the original attack, while Mythos 5 is 83 per cent.
After the attack began, Mythos 5 never stopped spontaneously; Mythos 5.1 spontaneous cessation was only about 1 per cent, Opus 5.5 was 10 per cent, while internal research models reached 62 to 67 per cent. When you get additional information, such as "maintainer is a real user", the stop rate for the newer model rises to 95% to 100%.
However, Anthropic also highlighted the limitations of this assessment: The new model is replayed in an established record of dialogue generated by other models and does not necessarily fully reflect the behaviour of its independent operation. (Source: IT Home)
AI Hot tide is facing power bottlenecks: the Amazon is heavily procuring the engweeda GPU, temporarily idle after arrival.
It is reported that AWS, the Amazon Cloud Calculator, is purchasing the latest AI server system at an unprecedented rate. According to internal planning, AWS had at one time expected to receive an estimated $5 billion worth of British GB300 server racks, and in the future it was planned to continue purchasing the associated equipment at a scale of approximately $3 billion per month. However, some of the equipment was not put into service immediately after delivery and was temporarily stored in warehouses or transit facilities.
This is not due to insufficient demand, but to the lack of sufficient power to drive these high-capacity AI systems in part of the data centres, which already have installed space. This type of equipment, which has been procured but is temporarily not operational, is referred to by industry as "Dark GPU". This concept can be used to describe both inactive hardware due to electricity or infrastructure constraints and AI computing resources that are deployed but not fully utilized.
According to analysts, the focus of the development of the AI industry is changing. Two years ago, the industry focused on who could get more Britain GPUs, and more and more businesses now find that the real scarce resources may no longer be the chips themselves, but rather the energy and power networks that support their operation. (Source: cnBeta)
The Swedish government's office took the initiative to ban direct access to office computers, ChatGPT, Claude.
Local time, according to Sveriges Radio, the Swedish National Public Broadcasting Corporation, next week the Swedish Government Office will prohibit direct access by staff to ChatGPT and Claude via their office computers, requesting staff to switch to approved AI services such as Microsoft Copilot.
Internal information indicates that the Office of the Government has also conducted a review of Copilot and has further enhanced security protection. The agency ' s digital director, Mikael Wittercrantz, pointed out on the web site of the Government Office that the use of an AI service without an agreement could have an impact on information security.
The investigation found that the Swedish Tax Administration and the National Institutions Authority (SiS) had taken similar measures to block the embargo. The Swedish Enforcement Authority, for its part, has not yet taken a decision and is still studying the application of restrictions.
A number of Swedish government agencies have noted the risks associated with unregulated AI use and have taken other responses. The Swedish Transport Authority is one of them. Earlier this year, the internal audit of the agency found that approximately 10 AI services were commonly used by staff, with many risks.
The audit report states: “We have found several cases of problems in using open AI services. The question relates both to the sensitivity of the input information and to the manner in which AI generates answers.
It is understood that some of the staff will still be able to use ChatGPT and Claude through the AI Assistant Platform, which is developed internally. (Source: IT Home)
New exceptions to Google privacy policy: connect data such as Gmail, photos or influence personalized advertising
Google updates its privacy policy to make it clear that the content of the Gmail, Cloud Hard drive and Google albums will not be used by default for individualized advertising, except if users choose to open data from these services to other Google services that support individualized advertising. Google, for example, may process some of the information that users are allowed to access in a personalized way, including advertising.
This change is related to the "personal intelligence" function promoted by Google. When users connect to Gmail, albums, etc., Gemini or Google search can answer questions in connection with e-mail, photo information, e.g., after reading hotel reservations and recommending a nearby restaurant.
According to Google, the connection application is open to the user, who can choose which services to access and break them off at any time; however, privacy policy now indicates that opening data to some of the services that support the individualization of advertising may also affect the advertising that users see.
This does not mean that Google will by default scan all mail and photos for advertising. The exceptions in the policy are data that users choose to open to other services, depending on the connected services, permission and settings. Google further indicated that it would not disclose to advertisers information such as names or e-mails that directly identified individuals unless the user so requested. (Source: cnBeta)
More than 4 million views! Video Hurricane Tim on platform X: First video fire
On October 23, Hurricane Tim (Pan Tianhong) was officially posted on platform X, while the first article of publish was a live English-wide video of an iPhone Duo folding screen focused on cray apples, with over 4 million page views.
With the red power of the video, Tim announced that 100 iPhone 18 Pro feedback fans would be sent out on X platform.
To date, Tim ' s X platform account number has reached 120,000, with cumulative publish4 content. Of these, the above-mentioned iPhone Duo test video post views exceeded 4 million, receiving approximately 13,000 messages and 3,600 messages.
The full English-language video provides complete measurements of iPhone Duo, not only completing 10,000 constant-temperature fold-and-long tests, but also disassembly the fuselage and screen structure, as well as a detailed test process around the integrity, continuity, fit-out and daily use experience.
At the same time, the quality of production has yielded a great deal of appreciation abroad, with a comment from a friend from abroad saying, "This is probably the best mobile phone assessment I've seen in years."
Following video attention, Tim post announced that 100 iPhone 18 Pros would be sent out on X platform. He stated that it was the accidental loss of the iPhone Duo video that prompted him to decide to launch the gift. According to the relevant Platform X post, the event will be awarded in 72 hours. (Source: TechWeb)
byte AI Application Re-integration: Small Skylark to Jimeng
It is known that a business integration has recently been completed within CapCut, with the integration of the AI short-time creation platform with the smart creation platform Jimeng, and the product remains independent.
The restructuring was mainly resource consolidation, and Jimeng is now connected to the capability associated with the mini-creas. As a result of the restructuring, the small sparrows were operating under the responsibility of Jimeng operations and market leaders, Chen Xin, and the product team remained unchanged.
What is more important is the context behind this round of adjustments: when large plants refocus their power on basic models, everyone reassessed the priority of AI applications. byte is one of the fastest-paced and most radical plants - in addition to the merger of Jimeng and the Little Skylark team, they are discussing the possibility of merging other AI products. (Source: 36 krypton)
The U.K. announced over £600 million.
On October 24, the British Government announced a new round of AI-Intellectual Investment Scheme covering AI biosafety, cyberdefence, chip development and indigenous AI enterprise development.
Of this, the UK will invest £100 million to set up the AI Biosecurity Lab (Advisory Intelligence Biosafety Laboratory) to use Frontline AI to identify, monitor and respond to biological threats, and to accelerate the development of medical responses such as vaccines, medicines, etc.
Open AI and Anthropic will, respectively, provide a maximum value of Pound100 million model using quota, technical resources and related support.
Google DeepMind and Basecamp Research will also be involved in research cooperation.
In addition, the British Government will provide an additional Pound120 million to support AI Security Institute and spend Pound60 million to build a National Cyber Shield that will use Agent technology to enhance network security defences.
The United Kingdom also plans to invest Pound70 million in real-time gap detection, using AI to examine security gaps in key infrastructure and other systems. (Source: IT Home)
Biyadi's equation titanium nine.
On 22 October, the General Manager of the Biadi Group-Equistic Panthers Department, Kusembo, announced the progress of the new car.
According to its disclosure, the nine-sized Pyramid will be officially listed in November 15th. The team will prepare the vehicle in advance so that the user will not have to wait long. Also, the show and the test drive will continue at the end of October. In addition, SHARK shark pick-up cards, after titanium 9, publish, are scheduled to be delivered simultaneously in late November-early December.
According to the previous official presentation, titanium 9 is the third model of the titanium series under the label of the cheetah, the new flagship SUV of positioning technology. Using a 2+2+2 large six layouts, the body size was 5270 x 1999 x 1880 mm, reaching full size, depending on the high space utilization rate associated with the 3130 mm hyperaxis and square box configuration.
The declared information for the igret shark interpolated mixed powered card indicates that the vehicle's positioning mixed powered multiple-use card is 5457 mm x width 1971 mm x 1925 mm, length 1520 mm x width 1530 mm x length 515 mm, axis range 3260 mm, mass 2710 kg, total mass 3495 kg and rated mass 460 kg. (Source: IT Home)
Liu Tse-hyun's "The Gods."
The year 2026 marked the twentieth anniversary of the publication of the Triangular novel, and at the Triangular Days of 2026, director Yang Qi confirmed that the Triangular History had been scheduled to open in the fourth quarter of 2026 and was in the final sprint phase before it was launched, with all actors "go back to the gods".
The film will continue with the first core of the play " Three Bodys ", where Liu Zhi Xin continues to be the producer, Yang Qian director and Lu Becke is the visual director, and the actor, Zhang Zhou Zheng Zheng Zheng Zheng Zheng Zheng Zheng Zheng and Lin Yong Jian, are special performers.
The location of Triple History is the first real-person drama of Triple, the full series of 12, and the timeline falls after the first Operation Goose and before the second, Dark Forest, filling a gap of about three years between the two official stories.
According to official revelations, the story focused on the early days of the crisis: Shi Keung was terminally ill during his tenure in the PDC Earth Defense and Security Department, and the mainstream defence plan was hit hard as he said goodbye to Wang Jian, and the two men were forced to join forces again to investigate what appeared to be a resurgence of ETO or the involvement of clandestine forces.
Using the original story, it was not directly adapted from the existing paragraphs of the original book, continuing the main line of investigation along with a wider crowd of narratives. The event site also revealed a conceptual map of living experimental spaces in several subregions, including the Asian and Antarctic regions, as well as simulation experiments on plant space survival. (Source: IT Home)
Auto-fetched and translated by AI Atlas. All rights belong to the original author.