Models cost millions to train and minutes to copy. Fine-tuning and distillation can erase where they came from.
Introducing the Fingerprinting Leaderboard, an open benchmark for model provenance — built so the proof lives in the weights, not the paperwork.
RT @opencommonspod: Closed models: you don't even know how big they are.
Open weights: you get the file. Not the data, not the training, not the licensing.
Open source: data, pipeline, and documentation all public.
Very few models clear that bar.
@dwddao on the difference.
Here are 4 fingerprinting schemes built with survival in mind:
> - implants a private key that causes the LLM to generate specific text when used
> - verifies ownership by cryptographically binding fingerprint prompts to their responses
> - embeds two orders of magnitude more fingerprints than existing schemes without model degradation
> - keeps fingerprints intact even after multiple models have been merged into one
Share
Copied
Sentient Foundation@sentient_found
Models cost millions to train and minutes to copy. Fine-tuning and distillation can erase where they came from.
Introducing the Fingerprinting Leaderboard, an open benchmark for model provenance — built so the proof lives in the weights, not the paperwork.
For Sentient Foundation’s Director of Venture and Growth @0xsachi, “Blaming the AI does not resolve the underlying failure in supervision or controls.”
@Inc breaks down why founders need to rethink how much authority they give AI agents.
Read more:
Last week, @sentient_found and @stripe hosted a VIP dinner in Seoul covering agentic commerce and model fingerprinting.
Shoutout to the @ScomotTeam and our Korean BD @namyura_ for helping plan this packed event!
RT @0xsachi: Science says data should be free and open to everyone.
For the communities @GainForestNow works with, data about nature is personal. Where the sacred place is. What one tree is used for.
Publishing it doesn't help them, and for centuries researchers have taken their data and given nothing in return.
Full episode on @opencommonspod:
Distillation attacks against AI companies are making headlines, but that’s not stopping Meta, DeepSeek, Liquid AI, and NVIDIA from dropping groundbreaking open-weight models.
Distillation attacks against AI companies are making headlines, but that’s not stopping Meta, DeepSeek, Liquid AI, and NVIDIA from dropping groundbreaking open-weight models.
LATEST: @deepseek_ai has released a new open-source model reportedly using 20× less compute than frontier models, as @Meta, @liquidai , and @nvidia accelerate open-weight releases.
Open-weight models accounted for 72% of tokens processed last week, while most spending remained concentrated on closed models. Could DeepSeek’s latest release start to shift that balance?
Share
Copied
Sentient Foundation@sentient_found
Distillation attacks against AI companies are making headlines, but that’s not stopping Meta, DeepSeek, Liquid AI, and NVIDIA from dropping groundbreaking open-weight models.
Want to top our Fingerprinting Leaderboard? Your ranking depends on three criteria, but here are the main ones you should pay attention to:
Robustness measures how well your fingerprint survives distillation, fine-tuning, and copying.
Utility measures how capable your model still is after it’s been fingerprinted.
Models cost millions to train and minutes to copy. Fine-tuning and distillation can erase where they came from.
Introducing the Fingerprinting Leaderboard, an open benchmark for model provenance — built so the proof lives in the weights, not the paperwork.
Verification, the final criterion, measures how reliably your fingerprint can be detected when the model is asked a secret question.
Find out where your scheme sits on
RT @0xsachi: Imagine you're negotiating on behalf of an African country, and the AI you're using to prepare was built on Western values.
@dwddao's question: how is that decision sovereign to you or your country?
Full episode on @opencommonspod :
As LLM copying and distillation becomes more common, our Fingerprinting Leaderboard aims to build an ecosystem where models can be openly shared while preserving attribution and ownership.
@Investingcom explores this approach with @SentientAGI researcher @edoardocontente.
Learn more:
Models cost millions to train and minutes to copy. Fine-tuning and distillation can erase where they came from.
Introducing the Fingerprinting Leaderboard, an open benchmark for model provenance — built so the proof lives in the weights, not the paperwork.
From working on self-driving cars to protecting rainforests, Dr. @dwddao of @GainForestNow has seen firsthand what AI is capable of.
Hear why he thinks AI works best as public infrastructure for the commons, rather than a tool for the rich, on the latest @opencommonspod ↓
Share
Copied
Sachi Kamiya 幸@0xsachi
A mayor in the Philippines cut down a protected mangrove forest to build a parking lot. When the community confronted him, he said it had always looked like that.
A year earlier, @GainForestNow had measured every tree there.
2,567 were missing.
The community took the data to court, and the mangroves were replanted.
My conversation with @dwddao, co-founder and chief scientist of GainForest is live on @opencommonspod.
We get into who should own nature's data, AI that helps people govern, and what open source really requires.
A mayor in the Philippines cut down a protected mangrove forest to build a parking lot. When the community confronted him, he said it had always looked like that.
A year earlier, @GainForestNow had measured every tree there.
2,567 were missing.
The community took the data to court, and the mangroves were replanted.
My conversation with @dwddao, co-founder and chief scientist of GainForest is live on @opencommonspod.
We get into who should own nature's data, AI that helps people govern, and what open source really requires.
Believe it or not, our Fingerprinting Leaderboard has been a year in the making.
See where it all began with @sentientagi’s @NeurIPSConf-accepted paper on OML ↓
Models cost millions to train and minutes to copy. Fine-tuning and distillation can erase where they came from.
Introducing the Fingerprinting Leaderboard, an open benchmark for model provenance — built so the proof lives in the weights, not the paperwork.
Our SF Tech Week happy hour at Novela, led by our Director of Events & Community @its_mils_, was one for the books 📚
Thanks to @WithAmpersand, @rootlyhq, @gauge_sh, and all the builders who stopped by to share notes on enterprise agents and open-source AGI!
Builders, don't forget our happy hour tomorrow 🥂
Join @WithAmpersand, @rootlyhq, and @sentient_found for an evening with the people building enterprise agents and open AGI.
📍 Novela, 662 Mission St, San Francisco
🗓 Thur, Oct 8th, 5:30-9:00PM PT
Builders, don't forget our happy hour tomorrow 🥂
Join @WithAmpersand, @rootlyhq, and @sentient_found for an evening with the people building enterprise agents and open AGI.
📍 Novela, 662 Mission St, San Francisco
🗓 Thur, Oct 8th, 5:30-9:00PM PT