Brave Leo’s linked posts claim that chats using GLM 5.3 Flash are encrypted in the browser and decrypted only inside an attested hardware enclave on NEAR AI Cloud. Brave says the browser encrypts each prompt before it leaves the device; NEAR Protocol says its TEE uses Intel TDX and NVIDIA confidential computing, and that Brave’s relay strips the user’s IP before the request reaches NEAR AI (Brave, NEAR Protocol).

Brave Leo’s linked posts claim that chats using GLM 5.3 Flash are encrypted in the browser and decrypted only inside an attested hardware enclave on NEAR AI Cloud. Brave says the browser encrypts each prompt before it leaves the device; NEAR Protocol says its TEE uses Intel TDX and NVIDIA confidential computing, and that Brave’s relay strips the user’s IP before the request reaches NEAR AI (Brave, NEAR Protocol).

The posts say Brave, NEAR AI, and the cloud host cannot read chat contents. NEAR AI adds that Brave checks the enclave attestation before sending a prompt. The announcement describes the privacy design; it does not independently establish that the claims have been verified in use (NEAR Protocol, NEAR AI).

Brave says the feature is available in Nightly, Beta, and Release. To use it, select “GLM 5.3 Flash” in Leo (availability and setup).