Erik Voorhees’ Venice has rolled out new encrypted AI interface models designed to make privacy verifiable rather than trust-based.
The update adds Trusted Execution Environment (TEE) and End-to-End Encrypted (E2EE) options alongside its existing anonymous proxy access and zero-data-retention processing.
TEE, operated by Venice’s external partners NEAR AI Cloud and Phala Network, runs AI workloads inside hardware-secured enclaves with attestation, preventing access by operators.
The company says remote attestation produces cryptographic proof that models are executing inside genuine enclaves, allowing independent verification of integrity.
While TEE prevents GPU operators and partners from accessing prompts during processing, data still passes through Venice’s proxy in standard encrypted transit. That means users continue to rely on Venice’s zero-retention assurances for data in transit, though they gain hardware-level guarantees at the compute layer, the team explains.
E2EE, meanwhile, keeps user prompts encrypted from device to GPU, decrypting only within verified environments.
Venice says neither the firm nor its infrastructure partners can access plaintext data at any stage, with each response accompanied by verifiable attestation evidence. E2EE is also supported by NEAR AI Cloud and Phala Network.
The trade-off is reduced functionality, as features like web search and memory require access to unencrypted data and are therefore disabled, as noted by the team.
TEE and E2EE are available only to Venice Pro subscribers.
Venice’s VVV token jumped 10% following the rollout, rising from $5.4 to nearly $6, per CoinGecko.
Source: https://cryptobriefing.com/encrypted-ai-inference-venice-release/
