HashSmash Competition Tests AI Attacks on Core Cryptographic Hash Functions

Eigen Labs and Shielded Labs, the organization supporting the Zcash privacy protocol, have jointly launched HashSmash, an open multiplayer competition designed to probe whether artificial intelligence can meaningfully advance attacks against the hash functions that form the backbone of modern internet cryptography.

The competition, hosted on a platform called Yukon, allows any participant to direct AI agents against four widely deployed hash functions: SHA-256, SHA-3, BLAKE3, and Poseidon. All results generated through the competition are published openly and subjected to review by professional cryptographers, lending the initiative a degree of academic rigor unusual for crowd-sourced security research.

Read more Home Bancorp to Release Q3 2026 Earnings After Market Close →

The initiative arrives at a moment of heightened concern among security researchers about the long-term resilience of cryptographic infrastructure in the face of rapidly advancing machine-learning capabilities. By stress-testing algorithms that underpin everything from blockchain transactions to secure web communications, organizers are seeking to establish an early-warning baseline for how AI-driven techniques might reshape the threat landscape.

Hash functions are one-way mathematical operations critical to data integrity, digital signatures, and password storage across virtually every sector of the digital economy. A demonstrated weakness in any of the four targeted algorithms — even a partial one — could have broad implications for software security standards and prompt accelerated migration to alternative schemes.

The open, competitive format is intended to aggregate findings from a wide range of participants rather than relying solely on institutional research, potentially surfacing vulnerabilities faster than traditional peer-review cycles allow. Continue reading at Cryptocurrency.