The rapid expansion of AI infrastructure is creating challenges in energy consumption and electronic waste (e-waste) management. Data centres accounted for approximately 1.5% of global electricity consumption in 2024, with their energy demand projected to more than double by 2030. Where electricity generation relies on fossil fuels, this growth compounds greenhouse-gas emissions and complicates decarbonization pathways, hindering progress towards SDG 13 (climate action). Under aggressive growth scenarios, e-waste from generative AI could increase by nearly three orders of magnitude between 2020 and 2030, reaching an estimated five million tonnes. However, less than one quarter of e-waste is formally collected and recycled. Informal practices such as open burning, landfilling and uncontrolled dumping release hazardous substances, obstructing advances towards SDGs 12 (responsible consumption and production) and 15 (life on land). AI technology can also generate highly realistic but factually incorrect texts, images and videos, blurring the boundary between truth and fabrication. The resulting spread of misinformation risks the collapse of public trust in authentic information, dissolving the shared factual foundation essential to SDGs 3 (good health and well-being) and 4 (quality education). This challenge is compounded by the opacity and limited explainability of black-box AI algorithms, which fundamentally threatens SDG 16 (peace, justice and strong institutions).