What if making data open is no longer enough to ensure trust? The digital world faces a crisis of disinformation and AI hallucinations, demanding a new level of data integrity.
The Objective Observer Initiative (OOI) provides the framework, and the Provenance Sandbox is the experimental data infrastructure that puts it into practice. This portal, built on specialized CKAN instances, addresses the critical challenge of ensuring AI safety by hosting Machine Learning datasets with complete and verifiable provenance. By implementing OOI’s "statements of intent" and active co-construction models, the sandbox shifts the role of a data portal from a passive repository to an active, accountable, and transparent public infrastructure.
The presentation will showcase how CKAN was customized to track these relationships between data, intent, and observation. https://mldata.opendata.ai/en/
Why this matters
Join the webinar to see how a standard CKAN portal can be re-engineered into a specialized, verifiably honest data environment for Machine Learning and AI safety. Attendees will learn the crucial intersection between data provenance, and AI ethics.
We will show you how to move beyond basic metadata to implement complex relationship tracking (Statements of Intent) using extended CKAN capabilities.
You will walk away with a vision of how CKAN can support not just open data access, but a verifiable data ecosystem that builds social cohesion and trust.
In the presentation, Steven De Costa will cover:
- The OOI Challenge: Why passive data sharing is insufficient in the age of AI disruption and disinformation.
- The Core Philosophy: Unpacking "verifiable honesty" and the crucial "Statement of Intent."
- Introducing the Solution: A detailed walkthrough of The Provenance Sandbox portal.
- CKAN Architecture: How CKAN was customized and extended to manage specialized ML/AI datasets and provenance metadata.
- Scaling with Partners: Highlighting collaborations with organizations like MLCommons and Source Cooperative to build a large-scale, co-constructed data ecosystem.
- A Deep Dive: Demonstrating how "statements of intent" and provenance are mapped and observed within the sandbox interface.
Details
- Date: Wednesday, Aug. 19, 2026
- Time: 13:00 UTC
- How long: 60+ mins
- Language: English
- Reserve your spot here.
Google Meet joining info
- CLICK HERE to join with Google Meet
- Or dial: ‪(BG) +359 2 907 4153‬ PIN: ‪492 965 761‬#
- More phone numbers here
Meetings agenda
- Welcome and introductions – 10 min
- Presentation – ~20 min
- Q&A session – 15 min
- Community announcements and sharing (if any) – 2 min
- Closing – 3 min
You can have a look at the [CKAN Monthly Live] Meeting Doc. Feel free to add items and things you’d like to discuss to the agenda.
Stay Connected
Follow us on X, LinkedIn, and YouTube, and join our LinkedIn Group!Â
You can also visit our website and blog to learn more!
Our LinkedIn Page is brand new - follow us there and spread the word!
Don’t forget to share your experience by using #CKANMonthlyLive!
We look forward to seeing you all on CKAN Monthly Live!
