We consider a random access system of non-cooperative selfish transmitters with the individual objectives of jointly optimizing throughput rewards, energy and delay costs. Our goal is to evaluate the effects of malicious nodes that have the dual objectives of blocking the packet transmissions of the other selfish nodes as well as optimizing their individual performance measures. We assume saturated packet queues of infinite buffer capacities and consider a general multi-packet reception channel that allows packet captures in the presence of multiple simultaneous transmissions. We formulate a non-cooperative random access game of selecting individual probabilities of transmitting packets to a common receiver and derive the transmission strategies in non-cooperative Nash equilibrium depending on the throughput rewards, energy and delay costs. The analysis provides insights for optimal strategies to block random access of selfish nodes as well as optimal defense mechanisms against possible denial of service attacks of malicious nodes in medium access control layer of wireless networks. In addition, we compare the results with the cooperative equilibrium strategies that optimize the total system utility and present a pricing scheme to improve selfish operation. For distributed implementation, we formulate a repeated game of the best response strategy updates and also develop an adaptive heuristic based on channel feedback only, if the system parameters are not explicitly known at the individual transmitter nodes.