Marks on Nostr: The LLM is hosted on a secure enclave GPU as well. The code running in the enclave is ...
The LLM is hosted on a secure enclave GPU as well. The code running in the enclave is open-source for anyone to audit. Attestation documents allow the client to know it is talking to the version of server code that is published online.
Lots more detail here:
https://blog.opensecret.cloud/opensecret-technicals/Live attestation proof here:
https://trymaple.ai/proof If I upload sensitive data like contracts, names, and financials, how do I know it's not getting leaked or compromised? I know you're using secure enclaves to process it, but how do you ensure your large language model isn't memorizing or storing this data? Can you give us a breakdown of how you're using secure enclaves to keep our data safe?
mapleai (nprofile…s7ns)
Published at
2025-05-20 15:24:21Event JSON
{
"id": "47159e089820b9fbb5585e750b430be3f520c2c0807d877217d560bdd2cfa924",
"pubkey": "8ea485266b2285463b13bf835907161c22bb3da1e652b443db14f9cee6720a43",
"created_at": 1747754661,
"kind": 1,
"tags": [
[
"e",
"84c29e44a8378a3940f3bd3b9ebd4b9d3189485d7714833cf452a3b9640363d8",
"",
"mention"
]
],
"content": "The LLM is hosted on a secure enclave GPU as well. The code running in the enclave is open-source for anyone to audit. Attestation documents allow the client to know it is talking to the version of server code that is published online.\n\nLots more detail here: https://blog.opensecret.cloud/opensecret-technicals/\n\nLive attestation proof here: https://trymaple.ai/proof\n\nnostr:nevent1qvzqqqqqqypzpgc6zlt80rf64slj6pk9yz22r8c9dnw8w47f730rvvu73v6w3ptwqqsgfs57gj5r0z3egrem6wu7h49e6vvffpwhw9yr8n699gaevspk8kqqtwgj8 ",
"sig": "10d155c3c43d1150c0bdf264f29cf4a40bfbfac666d55b981933f425074b6b32a0dd35d06cd170c01ed46caf801214706f86a4ef04a80bc651891b07bb139db1"
}
