This paper presents efficient distributed protocols for generating shares of random noise, secure against malicious participants. The goal is to create a distributed implementation of privacy-preserving statistical databases, where privacy is achieved by adding noise to query results. The noise is either Gaussian or exponentially distributed, and the protocols ensure that the results are accurate and secure. The paper introduces a distributed protocol called ODO (Our Data, Ourselves), which allows participants to generate shares of noise without a trusted server. The protocol involves sharing summands, verifying values, generating noise shares, summing all shares, and reconstructing the noisy sum. The main technical contribution is the cooperative generation of noise shares from Binomial and Poisson distributions, which approximate Gaussian and exponential distributions, respectively. The paper discusses the use of cryptographic tools such as verifiable secret sharing, non-malleable secret sharing, and secure function evaluation to ensure the security and correctness of the protocol. It also addresses the generation of Gaussian and exponential noise, showing how these can be approximated using Binomial and Poisson distributions. The results are shown to provide better accuracy and efficiency compared to non-interactive solutions like randomized response. The paper also discusses the generation of exponential noise using a shallow circuit that can generate arbitrarily biased coins at a low cost. The protocol is designed to handle Byzantine faults, assuming fewer than one-third of the participants are faulty. The results are applicable to a wide range of queries, including histograms and other low-sensitivity queries, and allow for individualized privacy policies based on the noise parameter R. The paper concludes that the ODO protocol provides a secure and efficient way to implement privacy-preserving statistical databases, allowing individuals to control the handling of their own information without relying on a trusted server.This paper presents efficient distributed protocols for generating shares of random noise, secure against malicious participants. The goal is to create a distributed implementation of privacy-preserving statistical databases, where privacy is achieved by adding noise to query results. The noise is either Gaussian or exponentially distributed, and the protocols ensure that the results are accurate and secure. The paper introduces a distributed protocol called ODO (Our Data, Ourselves), which allows participants to generate shares of noise without a trusted server. The protocol involves sharing summands, verifying values, generating noise shares, summing all shares, and reconstructing the noisy sum. The main technical contribution is the cooperative generation of noise shares from Binomial and Poisson distributions, which approximate Gaussian and exponential distributions, respectively. The paper discusses the use of cryptographic tools such as verifiable secret sharing, non-malleable secret sharing, and secure function evaluation to ensure the security and correctness of the protocol. It also addresses the generation of Gaussian and exponential noise, showing how these can be approximated using Binomial and Poisson distributions. The results are shown to provide better accuracy and efficiency compared to non-interactive solutions like randomized response. The paper also discusses the generation of exponential noise using a shallow circuit that can generate arbitrarily biased coins at a low cost. The protocol is designed to handle Byzantine faults, assuming fewer than one-third of the participants are faulty. The results are applicable to a wide range of queries, including histograms and other low-sensitivity queries, and allow for individualized privacy policies based on the noise parameter R. The paper concludes that the ODO protocol provides a secure and efficient way to implement privacy-preserving statistical databases, allowing individuals to control the handling of their own information without relying on a trusted server.