{"id":1296,"date":"2026-05-27T09:02:16","date_gmt":"2026-05-27T01:02:16","guid":{"rendered":"https:\/\/blog.liu-qi.cn\/2026\/05\/27\/x-daily-2026-05-26\/"},"modified":"2026-05-27T09:02:16","modified_gmt":"2026-05-27T01:02:16","slug":"x-daily-2026-05-26","status":"publish","type":"post","link":"https:\/\/en.blog.liu-qi.cn\/2026\/05\/27\/x-daily-2026-05-26\/","title":{"rendered":"X Platform May 26 AI Brief | Grok Build Public Beta Opens New AI Coding Frontier, Kimi Code Risk Control Upgrade Sparks Controversy, NetEase Youdao Open-Sources Full-Modal Model"},"content":{"rendered":"<h2 id=\"topic-13dbbedbe9\">Grok Build Public Beta: xAI Enters the AI Coding Arena<\/h2>\n<p>xAI officially released Grok Build, a terminal-oriented AI programming assistant competing with Claude Code and Codex. It supports two operating modes: interactive TUI full-screen and headless unattended (<code>grok -p<\/code>), featuring Plan mode, sub-agent parallelism, MCP external tool integration, Skills custom skills, and built-in Imagine image\/video generation. SuperGrok and X Premium+ users can install and use it directly. Elon Musk repeatedly promoted it, stating it is &#8220;still in beta but usable for production tasks.&#8221; Several bloggers had mixed reviews: some praised the TUI interface and fast X search, while others bluntly called it &#8220;garbage.&#8221; Notably, Grok Build supports switching to any custom model, not limited to Grok&#8217;s own.<\/p>\n<p>Sources:<\/p>\n<ul>\n<li>@xiaohu: <a href=\"https:\/\/x.com\/xiaohu\/status\/2059012699330424884\" target=\"_blank\" rel=\"noopener noreferrer\">https:\/\/x.com\/xiaohu\/status\/2059012699330424884<\/a><\/li>\n<li>@elonmusk: <a href=\"https:\/\/x.com\/elonmusk\/status\/2059016015644614797\" target=\"_blank\" rel=\"noopener noreferrer\">https:\/\/x.com\/elonmusk\/status\/2059016015644614797<\/a><\/li>\n<li>@Gorden_Sun: <a href=\"https:\/\/x.com\/Gorden_Sun\/status\/2059293612266192966\" target=\"_blank\" rel=\"noopener noreferrer\">https:\/\/x.com\/Gorden_Sun\/status\/2059293612266192966<\/a><\/li>\n<\/ul>\n<h2 id=\"topic-ff5a6fc1eb\">Kimi Code Upgrades Risk Control Strategy, Developer Community Reacts Strongly<\/h2>\n<p>Kimi Code announced an upgrade to its risk control strategy, restricting the relay use of Coding Plan. Many developers reported being affected, sparking widespread community discussion. Some pointed out that the Coding Plan adopts a &#8220;gym model&#8221;\u2014non-transferable\u2014but the risk control implementation feels awkward. Kimi&#8217;s official team later published an explanatory article in response to the controversy. A user shared the official explanation, noting that &#8220;today&#8217;s uproar is quite big.&#8221; This incident reflects the tension between platform providers and developer users in the commercialization of AI coding tools.<\/p>\n<p>Sources:<\/p>\n<ul>\n<li>@MaxForAI: <a href=\"https:\/\/x.com\/MaxForAI\/status\/2059291876327628864\" target=\"_blank\" rel=\"noopener noreferrer\">https:\/\/x.com\/MaxForAI\/status\/2059291876327628864<\/a><\/li>\n<li>@cellinlab: <a href=\"https:\/\/x.com\/cellinlab\/status\/2059267788267556964\" target=\"_blank\" rel=\"noopener noreferrer\">https:\/\/x.com\/cellinlab\/status\/2059267788267556964<\/a><\/li>\n<\/ul>\n<h2 id=\"topic-87568e0593\">NetEase Youdao Open-Sources Ziyue4 Full-Modal Model and TTS Engine<\/h2>\n<p>NetEase Youdao released and open-sourced Ziyue4, a 27B-parameter full-modal model achieving state-of-the-art (SOTA) in visual-mathematical tasks at the same scale, with a pure-text mathematical problem accuracy of 81.4%. The simultaneously open-sourced TTS model supports voice cloning with just 3 seconds of audio, covering 14 languages, with cloning accuracy exceeding 97% and timbre fidelity above 95%. Previously, NetEase Youdao had open-sourced the Ziyue-o1 lightweight reasoning model and the Ziyue3 mathematical model, designed to run on a single consumer-grade GPU. Based on the Ziyue model, they also developed products like Youdao LobsterAI and Youdao Simultaneous Interpretation Agent. This full-modal + TTS open-source release indicates Youdao&#8217;s expansion from education-specific models toward AI agent infrastructure.<\/p>\n<p>Sources:<\/p>\n<ul>\n<li>@vista8: <a href=\"https:\/\/x.com\/vista8\/status\/2059153797235228955\" target=\"_blank\" rel=\"noopener noreferrer\">https:\/\/x.com\/vista8\/status\/2059153797235228955<\/a><\/li>\n<li>@vista8: <a href=\"https:\/\/x.com\/vista8\/status\/2059154882133893562\" target=\"_blank\" rel=\"noopener noreferrer\">https:\/\/x.com\/vista8\/status\/2059154882133893562<\/a><\/li>\n<\/ul>\n<h2 id=\"topic-9377d8999d\">WeChat CLI Tool Users Receive Widespread Platform Warnings<\/h2>\n<p>Multiple bloggers reported receiving account warnings from Tencent after using the WeChat CLI tool. Analysis suggests the most likely trigger is the re-signing operation (<code>codesign --force --deep --sign<\/code>), which alters WeChat&#8217;s official signature, allowing the client to detect and report signature anomalies. Another possibility is the CLI tool reading local databases and keys during initialization. Vista8 warned users &#8220;never try it lightly&#8221; and destroyed related repositories in advance. Baoyu remarked, &#8220;WeChat is too closed; if it were more open, many interesting derivative products could emerge.&#8221; This incident once again highlights the tension between WeChat&#8217;s closed ecosystem and third-party tool development.<\/p>\n<p>Sources:<\/p>\n<ul>\n<li>@LufzzLiz: <a href=\"https:\/\/x.com\/LufzzLiz\/status\/2059210471371415654\" target=\"_blank\" rel=\"noopener noreferrer\">https:\/\/x.com\/LufzzLiz\/status\/2059210471371415654<\/a><\/li>\n<li>@vista8: <a href=\"https:\/\/x.com\/vista8\/status\/2059216142720512498\" target=\"_blank\" rel=\"noopener noreferrer\">https:\/\/x.com\/vista8\/status\/2059216142720512498<\/a><\/li>\n<li>@dotey: <a href=\"https:\/\/x.com\/dotey\/status\/2059093711875702898\" target=\"_blank\" rel=\"noopener noreferrer\">https:\/\/x.com\/dotey\/status\/2059093711875702898<\/a><\/li>\n<\/ul>\n<h2 id=\"topic-1b24143728\">Guizang Releases Xiaohongshu Image-Text Layout Skill with AI Auto-Matching and Format Adaptation<\/h2>\n<p>Guizang (op7418) released a Xiaohongshu image-text layout Skill based on HTML and real photos, which avoids being flagged as AI-generated content. The Skill automatically finds matching theme images from high-quality image websites, supports 3:4 magazine-style layouts, and can choose less intrusive formats to highlight original images based on image quality. When used with Codex, it automatically recognizes and separately labels spliced images. He also shared Swiss-style layout optimizations. This tool targets content creators on platforms like Xiaohongshu and WeChat Official Accounts, lowering the barrier for image-text layout.<\/p>\n<p>Sources:<\/p>\n<ul>\n<li>@op7418: <a href=\"https:\/\/x.com\/op7418\/status\/2059176154125480426\" target=\"_blank\" rel=\"noopener noreferrer\">https:\/\/x.com\/op7418\/status\/2059176154125480426<\/a><\/li>\n<li>@op7418: <a href=\"https:\/\/x.com\/op7418\/status\/2059258742831784053\" target=\"_blank\" rel=\"noopener noreferrer\">https:\/\/x.com\/op7418\/status\/2059258742831784053<\/a><\/li>\n<\/ul>\n<h2 id=\"topic-e37d2795ff\">AI Vibe Coding Practice Boom: Zero-Code Development Tools and Games<\/h2>\n<p>AI-assisted programming is spreading beyond the programmer circle. Imwsl90 shared that a friend with zero coding background used AI to create 10 tools and launched 7 apps with decent interface quality; one app generated 100,000 yuan in revenue from a single user. Gorden_Sun used GPT for character illustrations, Grok for video generation, and AI for coding to fully create an airplane shooting game deployed on GitHub with a detailed tutorial. Oran_ge observed a podcast claiming &#8220;now let AI propose requirements, develop, test, and deploy itself,&#8221; but also noted that &#8220;the real difficulty lies in the subsequent 100 versions.&#8221; These cases collectively reflect that AI coding tools are lowering the barrier to software creation, but iterative development remains a challenge.<\/p>\n<p>Sources:<\/p>\n<ul>\n<li>@imwsl90: <a href=\"https:\/\/x.com\/imwsl90\/status\/2059180971501736368\" target=\"_blank\" rel=\"noopener noreferrer\">https:\/\/x.com\/imwsl90\/status\/2059180971501736368<\/a><\/li>\n<li>@imwsl90: <a href=\"https:\/\/x.com\/imwsl90\/status\/2059206830505750591\" target=\"_blank\" rel=\"noopener noreferrer\">https:\/\/x.com\/imwsl90\/status\/2059206830505750591<\/a><\/li>\n<li>@Gorden_Sun: <a href=\"https:\/\/x.com\/Gorden_Sun\/status\/2059265886578729353\" target=\"_blank\" rel=\"noopener noreferrer\">https:\/\/x.com\/Gorden_Sun\/status\/2059265886578729353<\/a><\/li>\n<li>@oran_ge: <a href=\"https:\/\/x.com\/oran_ge\/status\/2059077685138411523\" target=\"_blank\" rel=\"noopener noreferrer\">https:\/\/x.com\/oran_ge\/status\/2059077685138411523<\/a><\/li>\n<\/ul>\n<p>Statistics: Scan timeline lines=360 Hit bloggers=32 Hit tweets total=182 Weighted tweet score=154.55 Original tweets=99 RT tweets=24 Fetch attempts=2 Boundary coverage status=tail_confidently_crossed_target_boundary<\/p>\n","protected":false},"excerpt":{"rendered":"<p>xAI launches Grok Build to enter the programming assistant field, Kimi Code&#8217;s risk control strategy adjustment sparks developer discussion, NetEase Youdao open-sources a full-modal model to expand AI infrastructure capabilities.<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[5],"tags":[19],"class_list":["post-1296","post","type-post","status-publish","format-standard","hentry","category-brief","tag-x--ai-"],"_links":{"self":[{"href":"https:\/\/en.blog.liu-qi.cn\/index.php\/wp-json\/wp\/v2\/posts\/1296","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/en.blog.liu-qi.cn\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/en.blog.liu-qi.cn\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/en.blog.liu-qi.cn\/index.php\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/en.blog.liu-qi.cn\/index.php\/wp-json\/wp\/v2\/comments?post=1296"}],"version-history":[{"count":0,"href":"https:\/\/en.blog.liu-qi.cn\/index.php\/wp-json\/wp\/v2\/posts\/1296\/revisions"}],"wp:attachment":[{"href":"https:\/\/en.blog.liu-qi.cn\/index.php\/wp-json\/wp\/v2\/media?parent=1296"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/en.blog.liu-qi.cn\/index.php\/wp-json\/wp\/v2\/categories?post=1296"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/en.blog.liu-qi.cn\/index.php\/wp-json\/wp\/v2\/tags?post=1296"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}