Eigen and Shielded Labs Launched AI HashSmash Competition
The project tests whether AI agents can identify vulnerabilities in cryptographic hash functions.
Updated on Oct. 5, 2026 in Artificial Intelligence
Live Poll
Do you trust that AI advancements are making the internet's cryptographic foundations more secure?
Eigen Labs and Shielded Labs launched HashSmash, an open competition evaluating AI agent performance against four cryptographic hash functions. Hosted on the Yukon platform, the initiative assesses whether current AI models can identify mathematical weaknesses in security protocols.
Why it matters
The competition aims to determine if AI can automate the discovery of cryptographic vulnerabilities, a capability that could redefine digital security standards. This research informs broader efforts, such as the Epoch project, to future-proof systems like Zcash.
The competition targets SHA-256, SHA-3, BLAKE3, and Poseidon hash functions. Participants utilize tools such as SAT solvers—algorithms for determining if a logical formula is satisfiable—automated differential search, and formal methods.
The players
Eigen Labs
A research entity focused on developing open AI testing platforms and cryptographic verification tools.
Shielded Labs
A laboratory focused on building cryptographically secure protocols, including the Epoch project for Zcash.
The details
Participants deploy AI agents to identify potential exploits within these four cryptographic standards. Every submission must clearly document the specific security mechanisms defeated and the precise computational resources required for the attack. An AI verifier combined with a committee of professional cryptographers reviews all findings to ensure they meet the competition's requirements. Results are made public to foster collaborative analysis of AI capabilities in security research.
Timeline
October 5, 2026: Launch of the HashSmash competition.
The Tech Race
This competition follows a pattern set by the Epoch project, utilizing crowdsourced intelligence to audit the long-term viability of current hash functions. It represents a shift toward stress-testing cryptographic standards against non-human automated analysis.
While the competition is primarily for researchers, the results will influence the security protocols integrated into future digital payment and identity systems. Security professionals and developers should monitor public results on the Yukon portal to understand how AI agents are changing cryptanalysis.
The takeaway
The project represents a crucial test of whether current AI can handle complex, logic-heavy cryptographic puzzles. Stakeholders should watch the Yukon portal for upcoming reports on whether agents successfully identify vulnerabilities in SHA-256 or BLAKE3.
Further reading
Learn more about evolving Artificial Intelligence research and testing benchmarks.
More information
Review competition requirements and public findings on the HashSmash competition portal.
Live Poll
Do you trust that AI advancements are making the internet's cryptographic foundations more secure?
