Researchers from Imperial College London have identified circuit layer vulnerabilities as the main threat to systems using Succinct Non-Interactive Arguments of Knowledge (SNARKs). The study analyzed 141 vulnerabilities across popular SNARK projects, revealing issues such as under-constrained, over-constrained, and computational errors. These vulnerabilities can compromise system soundness and completeness, with 95 affecting soundness and four affecting completeness. Developers face challenges in optimizing circuits for efficiency, impacting the cost of using SNARKs. Additionally, the conference highlighted Aptos team's implementation of weighted verifiable random functions (VRFs) to enhance consensus randomness. By incorporating weights based on stake, participants have varying probabilities of selection. Aptos successfully deployed this mechanism on its mainnet, providing unbiased and fast randomness. The distributed key generation process lasted approximately 20 seconds, showcasing significant progress in enhancing system security. Read more AI-generated news on: https://app.chaingpt.org/news