AI 快讯 / 日期归档2026-07-13
AI 快讯日报.
2026-07-13 收录的 20 条 AI 快讯,按采集时的来源记录回看。消息仍需核对原始发布页。
这期内容仍需回到原始发布页核对。
01快讯
Microsoft CEO Satya Nadella is calling out OpenAI and Anthropic for what he calls a "reverse information paradox." They train on public data under fair use but ban distillation of their own models, all while learning from customer interactions. Nadella wants companies to control their own learning infrastructure. Microsoft, of course, sells exactly that.
来源链接可在原站核对查看原文 ↗02快讯
阶跃终端品牌旗下首款智能体手机 STEPX Neo 今晚正式亮相,业内人士 @布鲁卡 今日发文称,这台是阶跃自己设计的手机,跟魅族倒是没啥关系,价格应该在 3.5K 往上。
来源链接可在原站核对查看原文 ↗03快讯
Some of the new features are powered by Google's Gemini AI assistant, which reflects the tech giant's broader push to integrate Gemini across its products while also better positioning Waze to compete with rival services such as Apple Maps.
来源链接可在原站核对查看原文 ↗04快讯
韩国全部人口中仍有约 1/3 无法使用人工智能,此外多数韩国用户使用的也是海外人工智能服务的免费版本。
来源链接可在原站核对查看原文 ↗05快讯
据用户反馈,豆包输入法鸿蒙版已开启招募,将带来豆包同款语音识别能力、越用越准的键盘输入和词语联想能力(暂仅支持 9 键、26 键输入模式)等。
来源链接可在原站核对查看原文 ↗06快讯
上海一男子利用 AI 技术伪造舌头被铁丝划伤的图片和视频,跨省对外卖商家进行恶意索赔,累计金额 3000 余元。其手法是通过外卖平台定位到千里之外的上海商家下单,再以“吃出异物”为由施压索赔。目前该男子已被警方采取刑事强制措施。
来源链接可在原站核对查看原文 ↗07快讯
Precursor, our new continuous behavioral validation engine for bot management, offers visibility into how humans and bots actually interact across the full user journey. By turning session-level behavior into bot detection signals, it identifies advanced automation with higher precision — while reducing friction for legitimate users.
来源链接可在原站核对查看原文 ↗08快讯
在阶跃星辰终端品牌发布会上,首款 AI 智能体手机 STEPX Neo 正式亮相。董事长印奇评价豆包手机是行业重要探索,将“智能机手机”概念带入大众视野,期待更多新品牌加速产品探索。
来源链接可在原站核对查看原文 ↗09快讯
阶跃星辰董事长印奇上回应为何涉足硬件领域,称要打造真正的大模型原生 AI 终端。首款智能体手机 STEPX Neo 亮相,配备交互背屏和 Step AOS 系统,并公布了美团、支付宝、百度等首批生态合作伙伴。具体配置和上市时间待定。
来源链接可在原站核对查看原文 ↗10快讯
我国首颗采用软件定义与三维近存计算技术的 AI 芯片亮相,通过底层架构创新,在 14 纳米制程上实现每秒 520 万亿次浮点运算,有效缓解“存储墙”瓶颈,并形成从加速卡到大规模智算集群的完整产品体系。这为人工智能算力底座提供了自主可控的新路径。
来源链接可在原站核对查看原文 ↗11快讯
阶跃终端品牌暨智能体战略发布会上,首款智能体手机 STEPX Neo 正式亮相,最大特点是机身背面配备了一块交互副屏。官方同时公布了包括美团、WPS、剪映、携程、高德、支付宝等在内的首批生态合作伙伴阵容。具体配置和上市时间待公布。
来源链接可在原站核对查看原文 ↗12快讯
The German consortium behind the AI model Soofi S has acknowledged in version 3.0 of its tech report that test questions from the science benchmark GPQA accidentally ended up in the training data. The community caught the error by examining the publicly available data. The team removed the benchmark from its evaluation and recalculated all results.
来源链接可在原站核对查看原文 ↗13快讯
当前环境异常,完成验证后即可继续访问。 去验证
来源链接可在原站核对查看原文 ↗14快讯
AI agent crawlers, the bots that fetch pages in real time on behalf of a person waiting for an answer, will be blocked by default on a slice of the web from September 15 onwards. Cloudflareannouncedthe change on July 1, and most of the coverage since then has focused on Google. The more useful part is what it asks of everyone building agents, and what it offers them in return.
来源链接可在原站核对查看原文 ↗15快讯
> I harass the sea with my tiny boat and am called a pirate, you do it with a great fleet and are called a king. Anthropic is actively campaigning to end software engineering. They need you to believe they can do that. Well, maybe it’s not you that they need to convince. Maybe it’s your C-Suite, various world leaders, or the manager of your retirement fund. They’ve raised $132 billion in investment, and are approaching an IPO valued over $1 trillion. Since they cannot show profitability, this depends on selling their hypothetical future impact.
来源链接可在原站核对查看原文 ↗16快讯
据博主 @Adak封狼居胥 分享,华为耳机联动豆包已全量上线,下载华为智慧音频和豆包 App,鸿蒙手机可以小艺 / 豆包双智能体在线,苹果 / 安卓也能通过豆包进行消息播报和面对面翻译。
来源链接可在原站核对查看原文 ↗17快讯
Waze announced several new features powered by Google’s Gemini, including conversational reporting and destination search.
来源链接可在原站核对查看原文 ↗18快讯
[](https://www.cookiebot.com/en/what-is-behind-powered-by-cookiebot/?utmsource=bannercb&utmmedium=referral&utmcontent=v2) Consent Details [[#IABV2SETTINGS#]](https://cacm.acm.org/news/can-we-understand-how-large-language-models-reason#) About
来源链接可在原站核对查看原文 ↗19快讯
微软宣布其 Windows 团队正全面利用 AI 挖掘系统漏洞,新引入的多模态 AI 安全系统 MDASH 能自动扫描并分析潜在漏洞,提升识别效率。这意味着未来的“周二补丁日”将包含更多修复内容,但并非漏洞增多,而是发现效率提高。
来源链接可在原站核对查看原文 ↗20快讯
Agentic LLMs often fail the same way, again and again. A Stanford research team traced this to missing, reusable capabilities. Their system, TRACE, diagnoses those gaps and trains for them directly.
来源链接可在原站核对查看原文 ↗