Generative AI is consuming vast amounts of water and energy, prompting calls for urgent action. A new US bill aims to address the environmental costs of AI, but much remains to be done. OpenAI's CEO, Sam Altman, has acknowledged the energy crisis facing the AI industry, warning that future AI systems will require significantly more power. He has invested in nuclear fusion, hoping it will provide a solution, though experts doubt its role in decarbonization by mid-century. Generative AI systems require large amounts of water for cooling and electricity generation. OpenAI's data centers in Iowa used 6% of the district's water in 2022, while Google and Microsoft saw significant increases in water use. Global demand for water for AI could reach that of the UK by 2027. The environmental impact of AI is a growing concern, with researchers and regulators urging more sustainable practices. The industry needs pragmatic solutions, such as using less energy, building more efficient models, and rethinking data center design. The BigScience project demonstrated that a model similar to GPT-3 can have a lower carbon footprint. However, current practices do not reflect this. Legislators are beginning to take notice, with a new US bill aiming to establish standards for assessing AI's environmental impact. However, voluntary measures are not enough. The AI industry, researchers, and legislators must work together to address the environmental impacts of AI. This includes measuring and publicly reporting energy and water use, prioritizing energy-efficient hardware and algorithms, and using renewable energy. Environmental audits and comprehensive reporting are also needed. Legislators should set benchmarks, incentivize renewable energy, and mandate environmental reporting. The Artificial Intelligence Environmental Impacts Act is a start, but more action is needed.Generative AI is consuming vast amounts of water and energy, prompting calls for urgent action. A new US bill aims to address the environmental costs of AI, but much remains to be done. OpenAI's CEO, Sam Altman, has acknowledged the energy crisis facing the AI industry, warning that future AI systems will require significantly more power. He has invested in nuclear fusion, hoping it will provide a solution, though experts doubt its role in decarbonization by mid-century. Generative AI systems require large amounts of water for cooling and electricity generation. OpenAI's data centers in Iowa used 6% of the district's water in 2022, while Google and Microsoft saw significant increases in water use. Global demand for water for AI could reach that of the UK by 2027. The environmental impact of AI is a growing concern, with researchers and regulators urging more sustainable practices. The industry needs pragmatic solutions, such as using less energy, building more efficient models, and rethinking data center design. The BigScience project demonstrated that a model similar to GPT-3 can have a lower carbon footprint. However, current practices do not reflect this. Legislators are beginning to take notice, with a new US bill aiming to establish standards for assessing AI's environmental impact. However, voluntary measures are not enough. The AI industry, researchers, and legislators must work together to address the environmental impacts of AI. This includes measuring and publicly reporting energy and water use, prioritizing energy-efficient hardware and algorithms, and using renewable energy. Environmental audits and comprehensive reporting are also needed. Legislators should set benchmarks, incentivize renewable energy, and mandate environmental reporting. The Artificial Intelligence Environmental Impacts Act is a start, but more action is needed.