Ссылка
click to show
click to show
There is an inherent trust problem in AI – it is called data sovereignty. There are policies in place that promise privacy, but practically no guardrails.
It is “trust me, bro” all over again. But what we need is “trust but verify”.
Solution: confidential computation in secure enclaves, verifiable, tamper-proof, and with reproducible attestation.
https://x.com/OasisProtocol/status/2082707772064551033