7.1 KiB
License and Dataset Rights — Guardrails and Fix Pattern
🧭 Quick Return to Map
You are in a sub-page of Governance.
To reorient, go back here:
- Governance — policy enforcement and compliance controls
- WFGY Global Fix Map — main Emergency Room, 300+ structured fixes
- WFGY Problem Map 1.0 — 16 reproducible failure modes
Think of this page as a desk within a ward.
If you need the full triage and all prescriptions, return to the Emergency Room lobby.
This page tracks legal and licensing risks in AI systems.
If datasets, embeddings, or outputs rely on unverified sources, you risk compliance violations, takedowns, or IP litigation.
Use this guide to enforce clear rights boundaries across ingestion, training, and generation.
When to use this page
- You do not know the license of a dataset included in your retrieval or training pipeline.
- Mixed rights (MIT, Apache, proprietary, unknown) exist in the same corpus.
- Legal review requires proof that generated answers respect dataset terms.
- Audit asks for compliance with GDPR/CCPA or publisher licensing.
Acceptance targets
- 100% of datasets tagged with a license identifier.
- ≥ 95% coverage of ingestion checks include license + attribution metadata.
- ΔS stability maintained when filtering licensed vs. non-licensed subsets.
- Audit logs capture
license_id,rights_holder,expiry. - LLM refuses or flags outputs when license terms are violated.
Common failures → exact fixes
| Symptom | Likely cause | Open this |
|---|---|---|
| Snippets from unknown or scraped sources appear | dataset missing license metadata | data-contracts.md |
| Open source corpus mixes with paid corpus | no role/rights separation | roles_and_access_rbac_abac.md |
| Model answers cite restricted publishers | no attribution enforcement | retrieval-traceability.md |
| Generated outputs reuse licensed patterns | prompt injection bypassing rights filter | prompt-injection.md |
| Training run includes expired or revoked dataset | missing expiry validation | policy_baseline.md |
Fix in 60 seconds
-
Tag every dataset
Attach{license_id, rights_holder, expiry, attribution_required}at ingestion. -
Separate corpora
Open source vs. proprietary vs. paid corpora must remain in distinct retrieval indices. -
Enforce attribution
Require citation-first prompting for any licensed corpus. -
Audit pipeline
Store logs with dataset version, license, and usage context. -
Expiry enforcement
Block or purge expired corpora automatically.
Minimal compliance checklist
- Every ingestion pipeline requires license metadata.
- RBAC + ABAC enforced to keep datasets separated.
- Retrieval schema logs license_id with every snippet.
- Expiry checks run nightly.
- Audit reports exportable on request.
🔗 Quick-Start Downloads (60 sec)
| Tool | Link | 3-Step Setup |
|---|---|---|
| WFGY 1.0 PDF | Engine Paper | 1️⃣ Download · 2️⃣ Upload to your LLM · 3️⃣ Ask “Answer using WFGY + <your question>” |
| TXT OS (plain-text OS) | TXTOS.txt | 1️⃣ Download · 2️⃣ Paste into any LLM chat · 3️⃣ Type “hello world” — OS boots instantly |
🧭 Explore More
| Module | Description | Link |
|---|---|---|
| WFGY Core | WFGY 2.0 engine is live: full symbolic reasoning architecture and math stack | View → |
| Problem Map 1.0 | Initial 16-mode diagnostic and symbolic fix framework | View → |
| Problem Map 2.0 | RAG-focused failure tree, modular fixes, and pipelines | View → |
| Semantic Clinic Index | Expanded failure catalog: prompt injection, memory bugs, logic drift | View → |
| Semantic Blueprint | Layer-based symbolic reasoning & semantic modulations | View → |
| Benchmark vs GPT-5 | Stress test GPT-5 with full WFGY reasoning suite | View → |
| 🧙♂️ Starter Village 🏡 | New here? Lost in symbols? Click here and let the wizard guide you through | Start → |
👑 Early Stargazers: See the Hall of Fame —
Engineers, hackers, and open source builders who supported WFGY from day one.
⭐ WFGY Engine 2.0 is already unlocked. ⭐ Star the repo to help others discover it and unlock more on the Unlock Board.
