EmbeddingGemma 2: an open, lightweight multimodal embedding model
deepmind.google·6d ago
AI Engineering
Lobsters·22d ago·5 min read
Adversarial examples for fast hash functions
TL;DR
Claude Fable's analysis of popular hash functions reveals that many perform poorly on certain inputs, with some hashes underperforming by at least 20 bits. This highlights the need for robust hashing methods that maintain speed without sacrificing security.
✦ Why It Matters
Evaluate your current hashing methods to ensure they meet security and performance standards.
Key Takeaways
- Assess the performance of your current hash functions against adversarial inputs.
- Consider using verified hashes to avoid potential vulnerabilities.
- Stay informed about the latest hash function analyses and improvements.
- Test your hashing methods with tools like SMhasher for empirical validation.
Full Summary
Hash functions convert data of varying lengths into fixed-size outputs, aiming to minimize collisions between distinct inputs. Fast hashing is crucial for applications like file synchronization, with xxHash achieving speeds of 60 GB/s.
However, many popular hashes, including komihash and HighwayHash, prioritize speed over robustness against adversarial inputs. Claude Fable utilized a tool to analyze these hashes, revealing that most exhibit significant weaknesses, performing at least 20 bits below expected collision resistance.
Some hashes have published proofs of their performance, but Fable identified errors in some proofs while verifying others using Lean, a formal proof verification system. This analysis underscores the importance of developing fast, reliable hash functions that can withstand potential attacks.