Does nostr:npub130mznv74rxs032peqym6g3wqavh472623mt3z5w73xq9r6qqdufs7ql29s fix this?
"Sousveillance (/suːˈveɪləns/ soo-VAY-lənss) is the recording of an activity by a member of the public, rather than a person or organisation in authority..."
Who's building the LittleSnitch of AI/LLMs?
https://www.obdev.at/products/littlesnitch/index.html nostr:note1a768p9kumsrnvmq7xxadweqyfr4pxsz352wxgx4lf2vf4s9vlkkq6xgyed
Discussion
Was wondering something similar. If nostr:npub130mznv74rxs032peqym6g3wqavh472623mt3z5w73xq9r6qqdufs7ql29s opens up a marketplace, can we find ways to validate not just the provider, but the actual models?
If someone claims they are serving llama-7-foobar could we find ways to have a signature scheme on the responses and who is the trusted validator?
I get that at some point you have to trust something, but I’ve been bothered by the idea that people are just claiming “here’s a model endpoint” and they could inject whatever the hell they want.
How do I verify my agent that depends on llama-7-foobar and your agent that relies on the same will be in agreement (nefarious or not doesn’t matter).
I think this is also a big part of what nostr:npub1mutnyacc9uc4t5mmxvpprwsauj5p2qxq95v4a9j0jxl8wnkfvuyque23vg is working to solve.