Injecting Undetectable Backdoors in Obfuscated Neural Networks and Language Models

When such backdoors exist, they allow the designer of the model to sell information on how to slightly perturb their input to change the outcome of the model. We develop a general strategy to plant backdoors to obfuscated neural networks, that satisfy the security properties of the celebrated notion of indistinguishability obfuscation . Applying obfuscation before releasing neural networks is a strategy that is well motivated to protect sensitive information of the external expert firm.