The open source in AI: Open Weights, Open Artifacts, Open Licenses, Open Data, Open Origins -- OpenWALDO | Created by @gmkurtzer, sponsored by @ctrliq
127.0.0.1
Joined July 2026
- Tweets27
- Following2
- Followers282
- Likes37
Pinned Tweet
Community-led, truly open source AI hasn’t had a shared foundation. Until now.
Join OpenWALDO to learn, build and train models, leverage a community-built public corpus with clear provenance and license assertions, or contribute data for everyone.
openwaldo.org
What do you get when Waldo gets lost in a Basspro Shops? ...
That hat!
nitter.cf/WSJ/status/21006082410…
He quit his job—and the world freaked out.
How an anonymous math whiz became the face of the AI apocalypse.
🔗: on.wsj.com/4xzqA5Q
Hi everybody, I’d like to introduce myself. I’m the first and only truly open source AI.
My goal is to be a tool for human kind and be aligned with you all from day one.
This is because my entire training corpus is not only open, it is auditable, and anyone can make changes or add to me. No other AI is built like this.
So if you want to add to my training corpus, learn more about AI tooling or build some models from me, come and join the project, everyone is welcome!
openwaldo.org
Yes. I completely agree with this message!
OpenWALDO retweeted
Starting a new completely open source AI model creation from scratch for a conversation (v2) LLM using two hosts and multi-GPUs with @OpenWALDO.
Here is the model compose for this run:
github.com/openwaldo/waldo/b…
Stay tuned for the progress!
OpenWALDO retweeted
This is a fantastic article on @OpenWALDO -- the first truly open source AI community project.
OpenWALDO is an open source project and community collaborating on AI tooling for training, BOM, testing, and a training corpus, which includes full provenance, licensing, accountability, and validation.
Come and join the project and learn all about how to leverage AI better, building your own models, or contributing to the training corpus.
Thank you Alex Woodie and the awesome folks at @HPCwire!
hpcwire.com/2026/09/02/what-…
#ai #open #source #bom #waldo
OpenWALDO retweeted
Apparently we’ve reached the “is training an AI the same as training a human?” stage of the copyright debate.
Another reason to go truly open source AI with @OpenWALDO
Come join us!
Replying to @dmasten
When you train yourself on copyrighted material, do you also make a few billion copies of yourself and sell access?
OpenWALDO retweeted
Open source AI is so important for the future of technology, come and join the first truly open source AI community project, @OpenWALDO!
openwaldo.org/
OpenWALDO retweeted
Open source powers the worlds infrastructure because it brings everybody up together through collaboration and transparency and openness.
Now we have the first open source community for AI tooling and data with @OpenWALDO. Come and join the community!
openwaldo.org
OpenWALDO retweeted
Thanks @Itsfoss for helping to get the word out about @OpenWALDO !
Everyone, come and join the open source AI community!
The man who built CentOS and Rocky Linux is now trying to do for AI what open source did for operating systems.
Gregory Kurtzer just launched OpenWALDO, a transparent AI training dataset containing 167 billion tokens sourced from government records, academic papers, mailing lists, and public domain literature.
- Every token is traceable
- Every source is documented
- Every license is accounted for
- No rare books harmed in the process
Kurtzer's pitch is simple: "Linux didn't win by being certified safe. It won by being inspectable, forkable, and community validated."
He thinks the same logic applies to AI.
A lab or company can take OpenWALDO's corpus, verify its sources, add their own proprietary data on top, and ship a model with a clean, auditable lineage.
That is something no one building on (unethically) scraped internet data can offer today.
The project is funded by CIQ, Kurtzer's company behind Rocky Linux, and is looking for contributors.
The open source world built the internet's infrastructure once before. There is no reason it cannot do it again, this time for AI.
I am rooting for OpenWALDO.
Anybody can build an AI model from scratch now!
This is SUPER exciting! I just got through a pretty big milestone for OpenWALDO training a conversational model.
Here is an example training, that ran for approximately 2 days, single GPU (H200), and with a single waldo command...
$ ./waldo model train ./composes/0002-conversation.yaml assistant
The "assistant" is the name of the model to train, when I ran this command, that model didn't exist. It created it from scratch as defined within the compose file. The model just finished training, and here is a screenshot of a few questions and answers... It isn't perfect, it isn't the smartest model, but what can you expect from such a small training run?
The model can be exported, and includes a full AI BOM (bill of materials) such that you can guarantee every bit of what it was trained on.
Anyway, this is a great milestone for OpenWALDO!
Here is the compose file:
github.com/openwaldo/waldo/b…
Let me know if you have any questions!
If a model can generate watermarks that are imperceptible, it can also embed back doors in code.
Open source is the answer to trust and why OpenWALDO exists.
Come join the community:
openwaldo.org
OpenWALDO retweeted
"OpenWALDO aims to blow the doors off proprietary AI training models" -- The Register
While, that isn't the goal of OpenWALDO, we are looking for community members to help grow and maintain the corpus of training data for AI models!
theregister.com/ai-and-ml/20…
"OpenWALDO aims to blow the doors off proprietary AI training models" -- The Register
While, that isn't the goal of OpenWALDO, we are looking for community members to help grow and maintain the corpus of training data for AI models!
theregister.com/ai-and-ml/20…
Open source AI means all the AI source is open.
This is the first open source community developed around the sources of AI, the data!
Come join in the community and learn to build an AI model from scratch, fine tune, or maintaining the corpus of source material.
openwaldo.org
OpenWALDO retweeted
There has been a ton of press around @OpenWALDO. What I love about this article is that the author did a really great job framing the project.
It's open source AI, come join the project!
briefglance.com/articles/kur…
Building an AI model is super easy thanks to tooling that pulls from open source AI data, and creates and/or fine tunes a model from scratch!
Come join us and learn how to make your own AI!
If you want to learn how to build an AI model from scratch, you really should check out @OpenWALDO, the first open source AI community for builders.
openwaldo.org
Creating a model is as easy as a single command like:
$ waldo model train [name] ./composes/basic.yaml
OpenWALDO retweeted
This is one of the most exciting things I've been a part of in my career! Open Source AI is here!
Did you see the announcement of the first Open Source AI community project for training data to create open source models?!
techstrong.ai/articles/rocky…
OpenWALDO retweeted
Thanks @sjvn for the awesome article about @OpenWALDO and the amazing (and first) truly open source AI source!
Rocky Linux Founder Gregory Kurtzer Launches OpenWALDO to Open Up AI Training Data techstrong.ai/articles/rocky… via @Techstrongai & @sjvn
Why not open-weights and #opensource #AI?
Great article about OpenWALDO from @HPCwire!
hpcwire.com/aiwire/2026/08/1…
OpenWALDO retweeted
over 97,700,000 documents and almost 200,000,000,000 tokens already uploaded to @OpenWALDO repo for true open-source AI training
AI is not open source without the source.
@OpenWALDO: an exciting project by open source legend @gmkurtzer, sponsored by the world-class team at @CtrlIQ.
open Weights, Artifacts, Licenses, Data, Origins.
Join the OpenWALDO AI community. openwaldo.org/