Harmful alcohol use is a major public health issue. In-person treatment has been hindered by the restrictions necessary during the Covid-19 pandemic. This study examined the effects of an at-home smartphone-based cognitive bias modification training in heavy drinkers. Experiment 1 tested the effect of a short 20-30-min smartphone-based approach-avoidance training (AAT) on image-induced craving at a 1-day follow-up. Sixty-two participants consuming 14+ units of alcohol/week were allocated to either the training or waitlist group. Experiment 2 used an updated version of the same short AAT intervention with a sample of n = 107 participants who consumed 20+ units of alcohol/week. Training effects at 1-week follow-up were compared to an active control group. Experiment 1 showed a significant reduction in image-induced craving for the training group at 1-day follow-up. Experiment 2 found that AUDIT weekly scores were significantly reduced at 1-week follow-up for the training group, all the while craving for soft drinks remained unchanged. Experiment 1 served as a first proof of concept for the efficacy of the new smartphone-based AAT training, and experiment 2 suggested that training effects on problem alcohol use hold at 1-week follow-up.