Aptos is consolidating three protocol-level confidentiality mechanisms on mainnet, giving users separate controls over identity, transaction amounts and pre-execution intent. According to Aptos Labs, Keyless Accounts, Confidential APT and Encrypted Mempool now form an opt-in privacy stack rather than a system that makes all Aptos activity private by default. Each feature protects a different category of information while preserving parts of the public blockchain record.
The newest component is Encrypted Mempool, which went live on October 12. Confidential APT had already reached mainnet earlier in 2026, while Keyless Accounts predate both features. Aptos is therefore packaging previously separate capabilities into a three-layer confidentiality model, not activating all three through a single new network upgrade. The architecture arrives as Aptos also expands institutional financial activity, including approximately $157 million of BlackRock BUIDL issued on the network.
Each Layer Protects Different Information
Confidential APT addresses the amounts layer. Users can move APT into the Confidential Asset system, where balances and transfer amounts are encrypted and zero-knowledge proofs demonstrate that transactions are valid without revealing those values. Sender and recipient addresses remain public. The feature provides amount confidentiality rather than anonymous payments, preserving an observable relationship between counterparties while concealing how much they hold or transfer.
The feature is available through Petra Wallet and Aptos Labs’ Confidential Assets Web interface. Aptos documentation also shows that confidential transfers require additional computation for proof generation and verification, making them more expensive than ordinary plaintext transfers. Confidentiality therefore remains an explicit user choice with additional execution overhead rather than the default behavior for every APT payment. This distinction matters on a network where stablecoin liquidity has already developed into a significant settlement layer.
Keyless Accounts protect a different boundary. They allow applications to derive blockchain accounts from credentials such as Google login without placing the user’s identifying OIDC information onchain. Zero-knowledge proofs demonstrate valid authorization while an ephemeral key signs transactions. Keyless reduces the public linkage between a login identity and its blockchain account, but it does not make transactions from that account invisible once they execute.
Encrypted Mempool Hides Intent Before Execution
Encrypted Mempool focuses on the period before a transaction executes. The client encrypts the function call and its arguments before submission, while the sender address and gas information remain visible. Validators order the encrypted payload and collectively decrypt it only after block ordering has finalized. The mechanism is designed to prevent traders or validators from reading actionable transaction intent early enough to front-run or sandwich it, rather than permanently concealing transaction history.
The feature uses threshold encryption tied to the Aptos validator set instead of a separate builder, TEE or external decryption committee. No individual validator can decrypt a pending payload alone, and Aptos says the system maintains the same stake-based trust assumption used for consensus. Petra exposes the option directly during signing, while developers can integrate it through TypeScript SDK v7.0.0 and later without modifying existing Move contracts. Network-level availability nevertheless does not mean every Aptos transaction is encrypted or that the feature has reached large-scale usage. The mechanism’s earlier mainnet activation established availability, not adoption.
Aptos has also identified Decibel as an application for encrypted order submission, where withholding order parameters until execution could reduce information leakage in onchain trading. That fits a network increasingly supporting financial assets such as native USDC and institutional settlement infrastructure. The important next metric is therefore usage: how frequently users, traders and applications opt into each confidentiality layer, rather than simply whether the protocol makes those tools available.