AI Pulse

Hugging Face开源可复用工具生成框架，让数据科学Agent自动生成任务特定工具，在DABStep基准达榜首。

Claude新增交互式图表、图解和可视化生成能力，提升数据展示和分析的交互体验。

Anthropic推进Claude在Excel和PowerPoint中的集成应用，增强办公生产力工具的AI辅助功能。

OpenAI Codex帮助Rakuten将软件交付速度翻倍，将故障恢复时间缩短50%并自动化CI/CD审查流程。

Rakuten uses Codex, the coding agent from OpenAI, to ship software faster and safer, reducing MTTR 50%, automating CI/CD reviews, and delivering full-stack builds in weeks.

OpenAI揭示ChatGPT Agent通过行为约束和数据隔离来防御Prompt注入和社工攻击的防护策略。

How ChatGPT defends against prompt injection and social engineering by constraining risky actions and protecting sensitive data in agent workflows.

OpenAI用Responses API、Shell工具和容器技术构建安全可扩展的Agent运行时，支持文件操作和工具调用。

How OpenAI built an agent runtime using the Responses API, shell tool, and hosted containers to run secure, scalable agents with files, tools, and state.

Wayfair用OpenAI模型优化电商支持和产品数据质量，自动化工单分流并增强数百万商品属性。

Wayfair uses OpenAI models to improve ecommerce support and product catalog accuracy, automating ticket triage and enhancing millions of product attributes at scale.

Hugging Face Hub新增Storage Buckets功能，为模型和数据集提供原生云存储解决方案。

Hugging Face分析16个开源强化学习库的最佳实践，分享Token管理和模型训练的优化经验。

Anthropic为Claude Code集成代码审查功能，增强AI编程助手的质量保证能力。

Google DeepMind回顾AlphaGo十年历程，展示其在科学发现和AGI探索中的深远影响。

Ten years since AlphaGo, we explore how it is catalyzing scientific discovery and paving a path to AGI.

IBM Granite 4.0推出1B参数轻量级多语言语音模型，为边缘设备优化部署成本。

Anthropic总结AI Agent工作流常见模式并指导选择，帮助开发者构建高效Agent系统。

Google推出Gemini 3.1 Flash-Lite，兼具极速推理和成本效益的小型多模态模型。

Gemini 3.1 Flash-Lite is our fastest and most cost-efficient Gemini 3 series model yet.

Google发布Nano Banana 2图像生成模型，结合专业级功能与闪电级速度的生产级方案。

Our latest image generation model offers advanced world knowledge, production ready specs, subject consistency and more, all at Flash speed.

Gemini 3.1 Pro面向复杂任务优化，提供深度推理和精准答案的高端多模态大模型。

3.1 Pro is designed for tasks where a simple answer isn’t enough.