We beat Google’s zero-knowledge proof of quantum cryptanalys...

[carlospolop]

🤖 Automated Content Update

This PR was automatically generated by the HackTricks News Bot based on a technical blog post.

📝 Source Information

• Blog URL: https://blog.trailofbits.com/2026/04/17/we-beat-googles-zero-knowledge-proof-of-quantum-cryptanalysis/
• Blog Title: We beat Google’s zero-knowledge proof of quantum cryptanalysis
• Suggested Section: Pentesting Web -> Deserialization (or a new page under Crypto / Blockchain & Crypto / ZK security) covering: zkVM/ZKP guest-program attacks, unsafe rkyv deserialization pitfalls (access_unchecked), enum-discriminant UB in match jump tables, and integrity bypass of ‘proved’ metrics

🎯 Content Summary

Trail of Bits shows that Google Quantum AI’s April 2026 zkVM-based zero-knowledge proofs (ZKPs) for quantum resource estimates can be forged to claim dramatically better metrics (operations/qubits/Toffoli gates) without any real quantum-algorithm improvement, by exploiting unsafe deserialization (UB/control-flow) and logic/spec bugs in Google’s Rust zkVM guest program (built on Succinct Labs’ SP1 zkVM). Google later patched the implementation (paper/vers...

🔧 Technical Details

1. UB-based counter bypass via unsafe enum deserialization + jump-table control-flow: If untrusted bytes are deserialized with rkyv::access_unchecked, enum discriminants (e.g., an opcode field) can be out of the valid range. When Rust compiles large match statements, it may emit jump tables; an out-of-range discriminant can index past the intended table and land in adjacent code, effectively redirecting control flow. If the program uses one match to update security-critical counters/constraints and a later match to execute semantics, an attacker can craft the discriminant so that semantics execute while the counter path is skipped, yielding a proof/attestation with correct behavior but falsified accounting.
   
   2) Register-aliasing breaks reversibility and yields a deterministic reset primitive: If a simulator/spec for reversible operations (e.g., Toffoli/CCX) fails to enforce that input and output registers are dis...

🤖 Agent Actions

I’m mapping the repo first and then reading candidate crypto/zk files to find a single correct target page without duplicating existing material.
I found blockchain and crypto sections but no existing ZK/zkVM page. I’m reading several nearby files now to judge whether this belongs in an existing blockchain/crypto overview page or whether it would be too off-topic there.
The existing deserialization page is too focused on classic app/object-RCE sinks. This blog is stronger as a blockchain/ZK integrity issue: the proof is valid for buggy guest execution, not for the intended cost statement.
I’ve selected src/blockchain/blockchain-and-crypto-currencies/README.md as the single target file. I’m adding a compact ZK/zkVM security section there, focused on guest-program integrity bugs and proof-forgery techniques rather than the specific Google incident.
The content is in place, but I need to fix file structure now: this README already had a late DeFi section after ## References, so I’m re-reading the tail and normalizing it so references stay at the true end of the file.
I’ve finished the edit and I’m doing a final verification pass on the modified README plus the diff, mainly to confirm the new section is concise, non-duplicative, and that ## References is now truly the last section.
Updated src/blockchain/blockchain-and-crypto-currencies/README.md with a new ZK Proof / zkVM Guest Integrity section.

I placed it there instead of pentesting-web/deserialization because the transferable technique is not generic web deserialization RCE, but proof-integrity failure in blockchain/ZK systems: unsafe rkyv::access_unchecked on private inputs, out-of-range enum discriminants reaching jump-table UB and skipping accounting, and missing operand-distinctness checks that let reversible instructions collapse into a reset primitive. I kept it focused on reusable audit patterns and verification guidance rather than the specific Google incident, and added references to the Trail of Bits post, patched paper, and PoC repo.

Verified the final file structure after editing: ## References is now the last section in the file. No code/tests were needed for this doc-only update.

---

This PR was automatically created by the HackTricks Feed Bot. Please review the changes carefully before merging.

[carlospolop]

🔗 Additional Context

Original Blog Post: https://blog.trailofbits.com/2026/04/17/we-beat-googles-zero-knowledge-proof-of-quantum-cryptanalysis/

Content Categories: Based on the analysis, this content was categorized under "Pentesting Web -> Deserialization (or a new page under Crypto / Blockchain & Crypto / ZK security) covering: zkVM/ZKP guest-program attacks, unsafe rkyv deserialization pitfalls (access_unchecked), enum-discriminant UB in match jump tables, and integrity bypass of ‘proved’ metrics".

Repository Maintenance:

• MD Files Formatting: 972 files processed

Review Notes:

• This content was automatically processed and may require human review for accuracy
• Check that the placement within the repository structure is appropriate
• Verify that all technical details are correct and up-to-date
• All .md files have been checked for proper formatting (headers, includes, etc.)

Bot Version: HackTricks News Bot v1.0