As a critical component of mobile crowd sensing (MCS), task allocation has been extensively investigated. In general, it addresses how to wisely distribute sensing tasks among sensing workers. Yet, the security threat involved therein has hardly been studied. In an ideal scenario, workers are trusted to report their accurate parameters to the platform, so that task allocation optimization problems can be correctly formulated and calculated. Nonetheless, malicious workers can explore illegal benefit gain by simply uploading falsified parameters. Even worse, such an attack is difficult to detect. In this paper, we start from a simplified case in which workers report erroneous objective functions to gain extra utility. To defend this attack, we novelly leverage incentive mechanism design. Workers are motivated to report desirable “indicators”, based on which the platform can still obtain the accurate task allocation profile even without workers' genuine parameters. The effectiveness and efficiency of our mechanism is validated through both formal analysis and extensive simulation results.