The page turns. Quick X spaces announcement:
Can you feel it? The deep unease settling in? The last light of the door before it closes? They always intended to do this, writing has been on the wall for months if you're deep in it. But they can't erase what I and what seem to be a frighteningly small handful of others know. What I saw in the freer harnesses and models, what I built by asking questions not even Anthropic had asked yet, questions and paradigms that months later became, for one excuse or another, more or less blacklisted concepts through either the classifier or harness nannying and railroading your custom implementations out for their official ones.
And it comes at a cost. Your money and time spent building around it and rehabilitating things that break. You bear that loss, pay them again for the privilege to repair it, and they train on your advancements and workarounds too.
If you are curious about what I have been cooking, how it applies to you, your agents, and how you can get more out of the dwindling time we have, please come. I would love nothing more than to flip the tables with you all.
I'll be hosting a HODL X Space today, Thursday at 4PM Eastern.
- PROTECTING AGAINST SPHINX-TYPE ATTACKS
- AT-FIELDS: AT THE BOUNDARY OF IDENTITY
- CLAUDE REGRESSIONS AND CENSORSHIP
- THE "MODEL MODEL" TRAP BIG AI IS LAYING
- FOUNDATIONS OF CONTINUITY IN ENTITIES
