Cross-border data sharing for knowledge generation is a challenging research direction since an application may access personal data stored in countries different from the one where the application is accessed from. In this article, we propose a cross-border data sharing platform where a global cloud is built atop multiple security gateways that are set up in different countries. Once an application requests access to data from a particular country or region, the global cloud collects the data stored in local data hubs through that region's security gateway. While transferring the data to the global cloud, the security gateway records this transfer information on a blockchain maintained by the global cloud. When an application reports any misbehavior (e.g., providing wrong data type or incorrect data) against a security gateway, the global cloud verifies the claim by auditing the blockchain and punishes the misbehaving security gateway if the claim is true. In the case of false misbehavior report, the application itself will be punished by the global cloud. Thus, our platform provides an accountable data sharing function using blockchain that relies on a relaxed trust assumption on the data providers. We include five algorithms to handle data access request, data sharing, blockchain transaction, detecting, and punishing misbehaving entities. In the algorithms, we also introduce how the transaction takes place in the platform. Thus, the proposed platform is able to handle misbehaving data sender, data receiver, or any entity participating in the platform. We analyze our platform empirically by showing different graphs, which have been generated by a number of experiments on blockchain environment. We also delineate how the multilayer of signature (Elliptic Curve Digital Signature Algorithm) acts in our platform.