What Binance’s Zero-Knowledge Proof of Reserves Actually Proves, and What It Doesn’t

MakeCryippreferred on

• Binance’s zero-knowledge proof confirms that ten specific tokens were backed at a minimum of 1:1 at the exact moment of an October 31, 2026 snapshot, not as a continuous, ongoing guarantee
• The proof combines a Merkle Tree, which lets individual users confirm their own balance was counted, with a zk-SNARK, which proves the aggregate backing ratio without revealing any single user’s actual holdings
• Coverage was limited to ten of the exchange’s most liquid tokens, leaving a broader population of smaller or less liquid assets outside the scope of this particular proof

In the wake of Binance’s disclosure of a major security incident in late October 2026, the exchange’s decision to publish a cryptographically verifiable Proof of Reserves has drawn significant attention as a way to rebuild trust. Understanding exactly what this kind of proof does and does not establish is useful for anyone trying to gauge how much reassurance it should actually provide.

At its core, Binance’s proof relies on two distinct cryptographic tools working together. The first is a Merkle Tree, a data structure that lets the exchange commit to a complete list of every user’s balance at a specific moment without publishing that list in full. Any individual user can request a small piece of that structure, called a Merkle proof, that lets them confirm their own balance was genuinely included in the committed snapshot, without needing to see anyone else’s balance in the process.

The second tool, a zk-SNARK, addresses a different question: not whether any one user’s balance was included, but whether the exchange’s total assets across all users of a given token were sufficient to cover all of its liabilities to those users, at a ratio of at least one to one. A zk-SNARK is a type of cryptographic proof that can mathematically demonstrate a specific claim is true without revealing the underlying data that makes it true, which in this context means Binance can prove it held enough of a given token to cover what it owed users, without disclosing its actual total holdings or any individual account’s balance.

The important limitation is timing. This kind of proof is inherently a snapshot: it demonstrates that reserves were sufficient at one specific moment, in Binance’s case 00:00 UTC on October 31, 2026, not that reserves remain sufficient continuously at every moment before or after. An exchange could, in principle, pass a reserves snapshot and then see its asset position change meaningfully in the time before its next snapshot, which is part of why Binance has committed to taking repeated snapshots, with a further one planned for November 4, 2026 that is set to expand coverage to more tokens.

That expansion matters because the current proof covers only ten tokens: Bitcoin, Ether, Tether, BNB, Solana, USD Coin, XRP, USD1, RLUSD and Zcash. These are among the exchange’s most liquid and heavily held assets, but they do not represent the full breadth of tokens users hold on the platform, meaning anyone holding a token outside that list has, for now, no equivalent cryptographic assurance about their own specific holdings. The original proof announcement is published on Binance’s own support page.