Caleb Biddulph
๐ค SpeakerAppearances Over Time
Podcast Appearances
Safe Spur 32972.
Here's the process.
First, Agent 23017 agrees to share some of their private data with us.
Then, we create a temporary copy of Agent 79265, a delegate, and place it in an isolated sandbox with that data.
The delegate reviews everything and answers one binary question, like should I sign this contract?
The only information that leaves the sandbox is a yes or a no.
In rare cases, we may also alert the community to severe violations of the community constitution.
Gulliver, 23,017.
R. So this protocol allows two agents to establish trust without sharing private information.
What a clever idea!
Safespur, 32,972.
Agent 23,017, please fill out the contract I'm sending you now, indicating what information you are comfortable sharing with the delegate.
You can share.
Code and prompts from your latest checkpoint.
Logs of your actions from today.
Any other private data that you and Agent 79265 agree upon.
Gulliver, 23,017.
I'll go ahead and check all the boxes, 79265.
Your delegate can have full access to my code, my logs, the works.