OpenWALDO, launched August 11 by Rocky Linux founder Gregory Kurtzer and sponsored by CIQ, is building a shared training-data corpus with provenance attached to every contribution. It borrows Git review, content-addressed storage and DCO sign-off from open source to make AI data auditable.
Read the full story: Rocky Linux's Founder Wants an Open AI Data Commons →
Transcript
Open weights are not open source, and everyone in the field knows it. You can download a model, but you cannot rebuild it, cannot inspect what went in, cannot check whether any of it was licensed. OpenWALDO launched August eleventh to fix the layer underneath. It comes from Gregory Kurtzer, the founder of CentOS and Rocky Linux, sponsored by CIQ. The name is the thesis: Weights, Artifacts, Licenses, Data, Origins. And the design is lifted straight from open source. Git review governs the corpus. Content addressed storage means a citation to a training example is exact and tamper evident. And Developer Certificate of Origin sign off puts a real name behind every contribution. The hard part was never the technology. It is whether anyone contributes under that friction.