Wildfires cause large damage to natural and human systems. Despite the clear connection between human-induced climate change and increased fire weather risk, a global, systematic attribution of observed extreme fires to human-induced climate change is lacking. Here, we address this gap by first linking observed regional weekly burned area extremes (>85th percentile) to the fire weather index (FWI) during the fire seasons of 2002–2015 via a logistic regression model, and then using simulations from climate models to quantify the impact of human-induced climate change. Focusing on regions with good predictability of the statistical model, we find that human-induced climate change was responsible for a fraction equal to 8% (±4%, standard deviation across climate models) of the predicted probability of more than 700 regional fire extremes on average, thereby increasing the probability of experiencing a fire extreme across 15 out of 19 analysed regions. While higher temperature is the main driver of the increased fire extreme probability, shifts in precipitation, relative humidity, and/or wind speed substantially modulated fire changes across many regions. Mainly because of warming, the probability of extreme fires attributable to human-induced climate change increased by 5.2%/decade globally over 2002–2015, in line with an acceleration of the climate-driven enhancement of fire extremes over the last decades that may continue in the near future. These findings highlight the urgent need for sustainable fire management strategies.