Canonical: https://builtonarc.app/guides/arc-privacy-features-for-institutions # Arc's privacy features for institutions, from Circle's own published material Updated 2026-09-02. Circle has published a detailed technical whitepaper on Arc's confidential-execution design. Its own docs still call the feature roadmap-only. > Circle has published a detailed technical whitepaper describing "Arc Privacy Sector" (APS), a confidential smart-contract execution system built on hardware enclaves, with real cryptographic design and performance figures. At the same time, Arc's own developer documentation states plainly, as of 2 Sep 2026: "Privacy features are on the roadmap and not yet available on Arc." Both statements come from Circle. This guide reports what each says and flags the tension between them rather than resolving it in Arc's favor. ## What Circle has actually published Two distinct Circle documents cover Arc privacy. The first is a dedicated privacy whitepaper, dated 8 June 2026, describing a system called Arc Privacy Sector (APS) — a confidential-computing design for private smart contract execution, presented in production-level technical detail with a performance benchmark section. The second is Arc's own developer documentation, a concepts page titled "opt-in privacy," which frames the same APS design but opens with an explicit status statement: privacy is not yet live on Arc. Both are genuinely Circle-authored and dated; Built on Arc did not find a third, independent source resolving which framing governs the actual mainnet-launch feature set. ## The mechanism, as described APS is designed as a second execution environment running alongside Arc's public EVM, inside hardware enclaves (the whitepaper specifies AWS Nitro Enclaves). A user encrypts a transaction to a network public key; validators running APS inside attested enclaves decrypt and execute it against private state, using ordinary EVM semantics. The public Arc ledger only ever sees an encrypted state-root commitment included in each block — from the outside, per the documentation, "this looks like an ordinary precompile call." Both the public and private ledgers are described as finalizing in the same block, under the same validator set and consensus round, which the whitepaper presents as the design's key differentiator over privacy systems that rely on separate chains or bridging. Disclosure is opt-in and default-deny: contracts choose, function by function, whether state is Open, Restricted, or Locked, and cross-contract access requires an explicit trust grant. Assets move between the public and private layers through a "shield/unshield" mechanism. Key material — a single master secret key — is split via threshold secret sharing across validators and only reconstructable inside attested enclaves, requiring the same 2t+1 threshold Arc's BFT consensus already uses for safety. ## Cryptography named The whitepaper names specific cryptographic building blocks: transaction encryption uses X-Wing KEM (combining X25519 and the post-quantum ML-KEM-768) with AES-256-GCM, and it reports private-transfer throughput of roughly 1,070 transactions per second against roughly 2,800 for public transactions in its own benchmark. These are stated, specific technical claims from Circle's own document — not third-party audited figures, and Built on Arc did not find an independent benchmark or security audit of APS published as of 2 Sep 2026. ## What the documents explicitly say is not solved The whitepaper itself lists real limitations rather than presenting the system as finished: it does not defend against side-channel attacks arising from enclave host memory-access patterns or timing; it does not hide contract bytecode, which stays visible on the public ledger; and it states that "designing private contracts whose ordinary query and execution paths do not leak the very state they are meant to protect is an open research problem." It also acknowledges the residual trust placed in the cloud hosting environment — an attacker with physical infrastructure access could, "in principle," observe encrypted-state write patterns. ## What institutions should take from the status tension Read narrowly, an institution evaluating Arc for confidential settlement should not assume private execution is available at mainnet launch on 16 September 2026 — Arc's own developer-facing documentation says the opposite, in plain language, as of 2 Sep 2026. Read separately, Circle has clearly done substantive engineering and design work on the feature, described in enough technical depth (named cryptography, a working benchmark, enumerated limitations) that this reads as more than a marketing roadmap slide. The honest summary is that Circle has designed and apparently tested a confidential-execution system it has not yet flipped on for Arc's production network, and has not published a date for when that changes. Sources: - [Arc Privacy Sector whitepaper (PDF)](https://6778953.fs1.hubspotusercontent-na1.net/hubfs/6778953/PDFs/Whitepapers/Arc_Privacy_Sector%20(5).pdf) - [Arc documentation: opt-in privacy](https://docs.arc.io/arc/concepts/opt-in-privacy.md) - [Arc.io: privacy whitepaper landing page](https://www.arc.io) *Built on Arc is an independent directory. Arc is a Circle product; we are not affiliated with, endorsed by, or operated by Circle.* ## Questions **Is Arc's privacy feature live at mainnet launch?** Arc's own documentation says no, describing privacy as "on the roadmap and not yet available" as of 2 Sep 2026 — despite a detailed whitepaper describing the design. **What is Arc Privacy Sector (APS)?** Circle's name for its confidential smart-contract execution design: a second, encrypted execution environment running inside hardware enclaves alongside Arc's public EVM, described in a whitepaper dated 8 June 2026. **Does APS use an encrypted mempool?** Not described that way. Transactions are individually encrypted to a network key before submission and decrypted only inside validator enclaves; the whitepaper does not use "encrypted mempool" as its own term. **Has APS been independently audited?** Not found. The performance and security claims in the whitepaper are Circle's own; Built on Arc did not locate a third-party audit as of 2 Sep 2026. **Is this the same thing as the post-quantum whitepaper?** No — they're separate documents. APS's encryption already uses post-quantum-hybrid cryptography (X-Wing KEM), but Arc's post-quantum roadmap covers a broader set of components (wallets, validators, infrastructure) on its own timeline.