Signal Brief

SpaceXAI 开源 Grok Build 并强化隐私保护

SpaceXAI 开源了 Grok Build harness CLI,并将数据保留功能默认设置为关闭状态。该版本实现了本地优先运行能力,允许用户使用自有推理后端,并已删除此前保留的所有编码数据,同时重置了所有用户的用量限制。

twitter关注列表 Elon Musk (@elonmusk) 发布 2026-07-15 收录 2026-07-16 观察

一句话判断

该更新提供了本地运行能力和默认零数据保留机制,适合关注 AI 编程工具隐私安全的开发者关注其开源实现。

核心信息

SpaceXAI 开源了 Grok Build harness CLI,并将数据保留功能默认设置为关闭状态。该版本实现了本地优先运行能力,允许用户使用自有推理后端,并已删除此前保留的所有编码数据,同时重置了所有用户的用量限制。

原始内容

Elon Musk (@elonmusk) 转发了 X Freeze (@XFreeze) 的帖子: Grok Build is now fully open source, and SpaceXAI has made its privacy-first approach even stronger SpaceXAI has officially: • Open-sourced the Grok Build harness CLI • Reset usage limits for all users • Disabled data retention by default • Deleted all previously retained coding data • Made the harness fully local-first, allowing you to run it with your own inference Since launch, Grok Build has supported Zero Data Retention (ZDR) for users who opted in Now, based on community feedback, SpaceXAI has gone a step further: Data retention is now OFF by default for everyone Your coding sessions remain private by default, with retention entirely under your control By combining an open-source harness, local-first execution, zero retention by default, and the deletion of previously retained coding data, SpaceXAI is setting a new standard for privacy in AI coding tools Transparent, open, and built to give users complete control over their data > **引用原帖 SpaceXAI (@SpaceXAI):** > We've open-sourced Grok Build and have reset usage limits for all users. > Open sourcing Grok Build allows anyone to support making a reliable and robust harness. Check out our code, including the Git repo for the Grok Build CLI. > https://t.co/3SSvPu2Nrz > https://x.com/SpaceXAI/status/2077494535387828644

相关动态

01

Mind Lab 开源长文本强化学习项目

Mind Lab 开源了一个长文本强化学习(RL)项目,支持 2M tokens 的长上下文。相比以往需要数千个 GPU 的 1M 上下文研究,该项目仅需 8 个 GPU 即可完成,大幅降低了超长上下文研究的算力门槛。

twitter关注列表2026-07-17#开源#技术突破#模型
观察
03

BestBlogs 早报 · 07-18

月之暗面发布Kimi K3,2.8万亿参数,896选16的Stable LatentMoE,上下文100万token,接近Fable-5但仍落后最强闭源模型,完整权重7月27日前开源;VentureBeat调查显示54%企业已发生AI代理安全事件;xAI开源Grok Build(84万行Rust代码)并残留上传用户代码痕迹;Cursor评...

twitter关注列表2026-07-17#AI#模型发布#开源
观察
05

Claude Managed Agents 支持500个技能项目升级

Anthropic 在Claude 3系列推出2024年2月新特性,实现代理系统技能数量扩展至500项,增强自定义化程度。更新后的API参数文档显示请求头需包含新版本控制标识符v20240201。相较Claude 2的100技能限制,新版本实现10倍扩容空间,核心优化体现在事件触发策略更新与上下文处理模块重构。

twitter关注列表2026-07-17#技术更新#产品改进#多模态
观察