Introduction Federated Learning (FL) empowers disseminated clients to collaboratively train models, while maintaining the privacy of their data. Yet, several security problems including malicious client behavior, poisoned data and model integrity compromise persist in classical FL systems. To this end, in this article, we propose Blockchain-enabled Federated Learning (BFL) system where we can deploy decentralized ledger technology for the tamper-evident model aggregation, transparent client contribution record and verifiable update validation. The designed BFL system uses smart contracts to enforce trust automatically without a need of intermediate party, Zero-Knowledge Proof (ZKPs) technique to keep the identity private with whom participates in activities, and tokenization incentive strategy that improve the reliability of proofs provided. It also presents a tunable blockchain overhead adaptive consensus algorithm to address scalability in various deployment environments including cybersecurity, healthcare, and Industrial Internet of Things (IIoT). Experimental results on benchmark datasets validate the efficiency of the proposed BFL framework by increasing 15-25% resistance to data poisoning and adversarial attacks associated to the state-of-the-art FL methods. The results also place BFL as a secure, transparent and scalable base to the future decentralized intelligent systems.