Production proposal adds a licensed research assistant and model approval conditions
The new production proposal prioritizes a research assistant for a bounded, licensed technical collection, with source evidence and agreed contributor payments. The project remains at specification stage.
Proposed first service
The proposal puts source permissions, attribution receipts and creator accounting before model adaptation. It separates citations, contractual payment rights and completed payments.
Model approval conditions
The proposed design retains the US-only and no-Chinese-model-lineage requirements. It keeps model routes disabled when required approval conditions fail or remain unresolved. It excludes production conversations and frontier responses from training unless specific reuse permissions exist.
OLMo ancestry findings
The review finds no disclosed Chinese pretrained-weight parent for OLMo-7B-0724-hf, but cannot certify every indirect dependency. It identifies proprietary synthetic content in training-data filtering. The evidence manifest records tokenizer comparisons and artifact hashes; the review excludes independent checks of weights and the full corpus.
Optional Palantir integration
The research proposes Palantir as an optional enterprise integration while retaining an independently controlled SIP core. The proposed integration leaves source permissions, model ancestry checks and creator compensation with the project. It calls for an agreement covering external customers, model training and independently owned SIP components.
Files in this change
| File | Change | Lines |
|---|---|---|
README.md | modified | +4 −1 |
docs/CHAT_HISTORY.md | modified | +14 −14 |
docs/OLMO_7B_0724_LINEAGE_RESEARCH.md | added | +131 −0 |
docs/PALANTIR_SOVEREIGN_AI_RESEARCH.md | added | +124 −0 |
docs/PRODUCTION_SYSTEM_PLAN.md | added | +225 −0 |
docs/SOLUTION_ARCHITECTURE.md | modified | +7 −5 |
docs/SOVEREIGN_SOURCE_AI_CODEX_SPEC.md | modified | +2 −0 |
docs/research/olmo-7b-0724-hf-evidence.json | added | +133 −0 |