01-12-Daily AI News Daily

AI News Daily 2026/1/12

AI News | Daily Morning Read | Aggregated Network Data | Frontier Science Exploration | Industry Free Voice | Open Source Innovation Power | AI and Human Future | Visit Web Version | Join Group Chat

Today’s Summary

Sakana & MIT collaborate on DRQ algorithm for AI programs to self-evolve through mutual predation
OpenAI officially supports Open Code, allowing direct authorization via ChatGPT
Tsinghua AGI Summit: Tang Jie declares the Chat era over, Lin Junyang gives China a 20% chance of overtaking
Top-tier products like DeepSeek originated as side projects
Open source projects ralph-claude-code and Plane project management tool gain traction

Product & Feature Updates

1. Sakana AI introduces a brutal training method where AI programs hunt each other in virtual machines. This novel approach is part of a Research Project (AI News) in collaboration with MIT. What’s really cool is the phenomenon of convergent evolution, where different programs actually evolve similar behaviors!

2. OpenAI has officially thrown its weight behind Open Code, allowing direct authorization via ChatGPT. This means Plus and Pro users can now confidently use Open Code (AI News) without fretting over account bans. Go ahead and call that Codex model to your heart’s content!
   

Frontier Research

1. The Digital Red Queen makes its grand entrance! Sakana dropped a super hardcore DRQ algorithm that forces AI programs to constantly evolve in Core War. What’s wild is that it led to convergent evolution, where different experiments ultimately developed Similar Strategies (AI News) . Oh, and the code is already open source, btw!

Industry Outlook & Social Impact

1. Tsinghua University played host to a powerhouse gathering of AI bigwigs, bringing together the “Four Grandmasters of Foundation Models” in Zhongguancun. The info density was absolutely wild! Tang Jie declared the Chat era pretty much over, while Yang Zhilin emphasized that building models is all about crafting a worldview. Lin Junyang even gave China a 20% chance of overtaking. This Full Report (AI News) is definitely worth a deep dive.

2. DeepSeek, a global top-tier product, actually started out as a side project! Just like how Fang Quan Quantitative’s side hustle became a worldwide sensation. Qwen kicked off as a side project too, and guess what? Claude Code followed suit. A Qubit Report (AI News) explored eight such cases. Makes you wonder, do side projects actually have a higher chance of striking gold?

3. Caution is needed against an anti-AI bubble, as open-source licensing disputes continue to heat up. The big question: Can GPL actually constrain model-generated content? Tailwind was even snagged as a case study for discussion. With code quality seemingly dropping and job anxiety spreading, this Detailed Discussion (AI News) offers a pretty diverse set of viewpoints. Let’s get pragmatic with experiments and not just blindly reject new tech!
   

Top Open Source Projects

1. ralph-claude-code has officially landed! This autonomous AI development cycle tool comes with smart exit detection, tailored specifically for Claude Code. It’s already snagged ⭐1.3k (AI News) stars on GitHub, and word on the street is the dev experience is super smooth.
   

2. Plane, a project management powerhouse, is here to replace Jira and Linear! This open-source gem covers absolutely all functions, from task sprints to document classification – it’s got it all. It boasts a whopping GitHub Stars ⭐43.4k (AI News) and sports a really slick modern interface design.
   

3. Twemoji, Twitter’s open-source emoji solution, is super handy and available to everyone! This emoji library has garnered quite a few ⭐17.4k (AI News) downloads and boasts excellent cross-platform compatibility.
   

Social Media Shares

1. Animated PPT generation is officially here! Guizang absolutely crushed it with animated presentation PPTs – seriously impressive stuff. It even supports exporting full videos, and the homepage animation loops infinitely. The Original Post (AI News) truly highlights a one-of-a-kind effect.
   
   

2. A solution for Claude Code login restrictions has surfaced! Turns out, modifying environment variables was no help; Anthropic played it pretty sly. But fear not – just add one line to the configuration file. Guizang shared the Specific Operation (AI News) : simply set hasCompletedOnboarding to true.
   
   

3. An innovative AI project installation prompt idea is making waves with oh-my-opencode! The cool trick? Installation instructions are actually written for the AI itself. Just copy the prompt, and boom — automatic installation! Guizang shared this Guizang’s Share (AI News) clever move, making Claude Code setup a one-click wonder.
   
   

4. Claude Code writing Claude Code – talk about a mind-bending scenario! We’ve got Agents writing Agents, and the project dependencies are seriously nested. To make things even wilder, functionality apparently broke after an upgrade! Nazha complained Nazha’s Complaint (AI News) about this “too magical” world.
   
   

5. Too much certainty leads to nothing – that’s the core of Fanren Xiaobei’s thought-provoking reflection. “Playing it safe is actually high risk,” they argued, emphasizing that innovation truly happens in those fuzzy, uncertain zones. A mere 20% success rate isn’t going to wait around for anyone! This Original Post Reflection (AI News) is pretty insightful, if you ask me.

AI News Daily Audio Version

🎙️ Xiaoyuzhou	📹 Douyin
Laisheng Tavern	Self-Media Account