Been digging into @OpenLedger for a few days now and one thing keeps pulling at me more than the rest.
The attribution engine update on January 26 the one that ensures data to output links stay intact even after a model gets fine tuned, that's the thing I keep coming back to.
Because most "fair AI" projects break at exactly that seam. Model gets updated, the original data trace gets diluted and the contributor stops seeing rewards. OpenLedger's Proof of Attribution is designed to persist through those updates.
Every inference event still traces back to the source data. Contributor gets paid not once, at training time, but continuously, as the model keeps running.
That's genuinely a different design from what I expected. I went in thinking this was mostly a data marketplace with a token bolted on. Turns out the economic mechanism lives at the inference layer, not the dataset upload layer. That's actually the hard part to get right.
Still… I don't know how many active datanets are running production inferences right now versus just being seeded for future use. The $OPEN 24h volume sitting around $24M is fine, but it doesn't tell you whether real AI developers are pulling attributed inferences or whether most of the chain activity is still just node operators and testnet habits carrying over.
Whether that inference level payout actually flows to anyone outside early contributors who set up datanets in the first place, that I haven't been able to verify cleanly yet.