Originally postulated in 2001, the impaired response inhibition and salience attribution (iRISA) model of addiction highlights the prefrontal cortex (especially the orbitofrontal, dorsolateral, anterior cingulate, and inferior frontal regions) as central to drug addiction symptomatology. Accordingly, drug cues assume a heightened salience and value that overpower alternative reinforcers, with a concomitant decrease in inhibitory control, especially in a drug-related context. These processes may manifest in metacognitive impairments (e.g., self-awareness of choice), obstructing insight into illness, as a function of recency of drug use. In this review, we update the neurobehavioral evidence for iRISA two decades later, emphasizing the robust measurement of the iRISA interaction (between a drug-related cue/context and a cognitive-behavioral function), and highlight relevant individual differences (e.g., drug use severity, craving). Crucially, we describe data suggesting functional recovery (with abstinence, treatment, and other emerging modalities) and the need for identifying valid outcome biomarkers. We end by highlighting recent developments in artificial intelligence (e.g., natural language processing applied to spontaneous speech) and computational modeling, and call for enhanced ecological validity to facilitate dynamic and clinically meaningful neural explorations in drug addiction.