On September 24, 2026, a research group at AmericanFortress published a paper called “Provenance Proofs.” It describes a way to prove that two addresses on different blockchains come from the same hidden wallet seed, without revealing the seed. The authors named in press coverage are Vincenzo Botta, Michal Pospieszalski, Emanuele Ragnoli, and Justus Ranvier.
The target is a bridge or exchange that must pay out to the right destination. If a deposit on Bitcoin and a payout address on Solana both link to the same seed, a substituted address fails the check. Today this is a paper with benchmarks. It is not part of a wallet or bridge you can use.
How the proof works, as described
According to HackerNoon’s write-up, the design has three forms. A joint proof covers a chosen set of addresses. A deferred proof uses randomized commitments. A persistent tag lets a verifier recognize repeat proofs from one wallet. In the Bitcoin to Solana case, the bridge checks the deposit, both derivation proofs, and the link between them before it releases funds. The paper also extends the check to refund addresses.
AmericanFortress CEO Michal Pospieszalski told Bitcoin.com News that an attacker would first need to commit equal capital to get past the check. That is his claim, and we found no independent evaluation of it.
Two sets of speed claims
The paper and the interviews do not line up. One is a measured benchmark from the authors, the other is a statement about a product that does not exist yet.
| Detail | Paper benchmarks (HackerNoon) | Interview statement (Bitcoin.com News) |
|---|---|---|
| Proving time | About 13.3 seconds for a Bitcoin to Solana bundle (two full-path proofs of about 6.7 seconds each) | Roughly three seconds with an SDK integration with MetaMask |
| Verification | About 475 ms per full-path proof | Not stated |
| Proof size | About 9.7 MB per full-path proof, about 19 MB for the two-chain bundle | Not stated |
| Status | Authors’ benchmarks on a 16-core machine with the Plonky3 proving system | A plan stated in an interview, not a release |
The three-second figure conflicts with the 13.3 second paper benchmark, and it describes a planned integration rather than something measured. The large proof sizes matter for any wallet or bridge that has to move them around.
Who Benefits and How
The intended beneficiaries are bridges and exchanges, and through them the people who use them. A working version could catch a payout sent to the wrong cross-chain destination before funds leave. Nothing is available to use yet: the sources describe a paper, benchmarks, and a planned MetaMask SDK integration. If it ships, bridge operators and wallet developers would feel it first.
Where it falls short
The linkage step is not benchmarked alone. HackerNoon says the combined circuits are not yet benchmarked. HackerNoon also says AmericanFortress launched a different product, SafeSend, in beta on Arbitrum in May, but it does not say provenance proofs are part of it. An exchange integration is described as a design sketch, and the work is covered by a pending patent application.
It also does not fix address poisoning on a single chain. That is the scam where someone plants a look-alike address in your history. Cointelegraph reports, citing onchain investigator Web3 Antivirus, that a user lost 49,999,950 USDt after copying a look-alike address from their transaction history. The first three and last four characters matched, according to SlowMist founder Cos. The victim had sent a small test transaction to the right address minutes earlier. The attacker then swapped the funds to Ether, split them across wallets, and moved part into Tornado Cash. The page we opened gave no date or chain, so we do not state either. Nothing we read says this proof would have stopped that transfer, because it checks cross-chain bridge payouts.
Common questions
Can I use this in my wallet today?
No. The sources describe a research paper and benchmarks. A MetaMask SDK integration is a plan stated in an interview.
Does it reveal my seed phrase?
By design, no. The paper’s goal is to prove two addresses share a seed without exposing the seed or the link between them. The design has not been independently reviewed in anything we found.
Would it stop address poisoning scams?
Not as described. The proof checks that a bridge deposit and its payout address belong to the same wallet. A same-chain transfer to a poisoned address is a different attack. Checking the full address before sending remains the practical defense.
Is it quantum resistant?
Bitcoin.com says the proofs are designed to work with post-quantum standards. The benchmarks in HackerNoon use secp256k1 keys. We could not verify the quantum claim.
Sources
- HackerNoon: AmericanFortress Publishes a Zero-Knowledge Proof That Stops Bridges Paying the Wrong Wallet (Sep 24, 2026)
- Bitcoin.com News: Zero-Knowledge Breakthrough Makes Bridge Hacks “Unviable” (Sep 28, 2026)
- Cointelegraph via TradingView: How a single copy-paste mistake cost a user $50M in USDT (accessed Oct 13, 2026)