之前发了很多期商业方向的博客了，其实我也是一个技术博主来着，而且研究AI，不研究技术，属于水中捞月。所以，本期我来一份硬核技术分享。

这个人叫 POM（Peter O'Malley），开源 AI 艺术社区 Banodoco 的创始人。他不仅放出了自己的全部对话数据，还开源了一个叫 DataClaw 的工具，帮所有人把自己的 Claude Code 和 Codex ...

Amazon Threat Intelligence has revealed that a single “unsophisticated” attacker has compromised more than 600 organisations across 55 countries in ju.

The module targets Claude Code, Claude Desktop, Cursor, Microsoft Visual Studio Code (VS Code) Continue, and Windsurf. It also harvests API keys for nine large language models (LLM) providers: ...

DeepSeek V4 Lite and GPT 5.3 (Garlic) represent two of the most-discussed developments in artificial intelligence this week, as overviewed by Universe of AI. DeepSeek V4 Lite, overviewedly leaked ...

OpenAI has accused DeepSeek of malpractice in developing the next version of its artificial intelligence model — even before any official launch. “DeepSeek’s next model (whatever its form) should be ...

Feb 12 (Reuters) - OpenAI has warned U.S. lawmakers that Chinese artificial intelligence startup DeepSeek is targeting the ChatGPT maker and the nation's leading AI companies to replicate models and ...

A new version of DeepSeek is reportedly better than both ChatGPT and Claude on coding. Tech investors have been growing concerned about sky-high spending on artificial intelligence. But the concern ...

Developers have identified references to an unidentified “MODEL1” in DeepSeek’s GitHub repository, suggesting preparations for a new flagship model. The discovery follows earlier reporting that ...

Jan 9 (Reuters) - Chinese AI startup DeepSeek is expected to launch its next-generation AI model V4, featuring strong coding capabilities, in mid-February, The Information reported on Friday citing ...

For the past few years, the recipe for building smarter artificial intelligence has been simple: make it bigger. Add more layers, feed it more data, and watch the intelligence emerge. That brute-force ...

DeepSeek introduced Manifold-Constrained Hyper-Connections (mHC) to improve large-model training scalability and efficiency. The mHC method was tested on 3B, 9B, and 27B parameter models, showing ...