Al Gore warns AI misuse poses greater risks than data centers

Al Gore told TechCrunch that he is not worried about the carbon footprint of AI data centers, but he is concerned about the industry’s own warnings about where the technology is headed. He said the biggest threat is not the energy used to train models, but the potential misuse of AI in areas like surveillance, disinformation and autonomous weapons. The former vice‑president and climate activist said he “doesn’t lose sleep over data‑center emissions.”

The comment comes as AI growth accelerates. In 2024, the world’s AI training workloads are projected to double, pushing data‑center electricity use up by an estimated 15 % annually. Climate experts say AI could add 0.3 % to global CO₂ emissions by 2030 if it relies on fossil‑fuel grids. Yet, Gore’s focus is on the social risks highlighted by AI researchers, such as algorithmic bias, privacy erosion and the spread of fake news. He argues that without safeguards, the technology could outpace our ability to manage its consequences.

During the interview, Gore referenced studies showing that training a single large language model can consume as much energy as five people’s yearly household usage. He noted that tech firms like OpenAI, Google and Microsoft are investing in renewable energy, but that many of the same companies are also developing tools that can be weaponised. Some industry leaders say climate is the bigger danger, while privacy advocates push for stricter oversight. Gore added that the industry’s warnings about “unintended harms” should be taken seriously, even if they are hard to quantify.

Looking ahead, Gore urges policymakers to set clear rules on AI deployment, transparency and accountability. He believes that without robust governance, the technology could undermine democratic institutions and exacerbate inequality. The debate is heating up as governments in the EU, US and China announce new AI regulations. If the tech sector does not act, society may face disruptions that outstrip the environmental impact of data‑center energy consumption.