The OML standard for open-weight AI



Closed captures value, open leaks it. Builders are stuck between control and reach

@SentientAGI brings a credible path with an academic backbone at NeurIPS 2025

OML = Open, Monetizable, Loyal

- Open weights and local inference
- Cryptographic authorization and per-query receipts
- Policy loyalty under white-box threat models

OML 1.0 shows it in practice

- 25,000 hidden fingerprints in LLMs
- Survive fine-tuning, merging, distillation
- Undetectable in normal use with no perf hit
- Trace copies and prove provenance

Survey of TEEs, obfuscation, homomorphic encryption, with proofs that perfect protection is impossible while practical economic-crypto guarantees hold

Authors across Princeton, UW, UIUC, Sentient Foundation, with Andrew Miller, IC3, and Teleport TEE expertise

Outcomes: licenseable open weights, revenue on verified use, portable constraints, a viable open-source AI market

Details:
This page may contain third-party content, which is provided for information purposes only (not representations/warranties) and should not be considered as an endorsement of its views by Gate, nor as financial or professional advice. See Disclaimer for details.
  • Reward
  • Comment
  • Repost
  • Share
Comment
0/400
No comments
Trade Crypto Anywhere Anytime
qrCode
Scan to download Gate App
Community
English
  • 简体中文
  • English
  • Tiếng Việt
  • 繁體中文
  • Español
  • Русский
  • Français (Afrique)
  • Português (Portugal)
  • Bahasa Indonesia
  • 日本語
  • بالعربية
  • Українська
  • Português (Brasil)