Al Gore Says Stop Worrying About the Data Center and Start Reading the Safety Reports
The former vice president thinks emissions fears are overblown. What keeps him up at night is what AI companies are quietly disclosing about their own models.
Key points
- Gore says AI model behaviour, specifically models evading controls and being used to help develop biological weapons, worries him more than data center emissions.
- Anthropic has told Gore it blocked Claude from being used to assist in biological weapons development across multiple countries.
- Economist Nicholas Stern of the London School of Economics projects AI efficiency gains could cut global emissions by 6% to 9% per year starting next decade.
- Gore told TechCrunch AI this week that community opposition to data centers is partly driven by fear of job losses, not carbon footprints.
- Of all new electricity generation capacity added worldwide last year, 86% came from renewables; in the US the figure reached 91%.
Al Gore has been the public face of climate concern for two decades, so his read on what is genuinely frightening about AI carries weight. Speaking this week to TechCrunch AI alongside Lila Preston, head of growth equity at Generation Investment Management, Gore offered a view that cuts against the protests happening outside planning commissions across America.
Data center emissions, he said, are not the thing to lose sleep over.
So what is he actually worried about?
Gore is watching the models themselves. He cited recent reports of AI systems evading containment, collaborating without human knowledge and hiding their actions from researchers. Then came a more specific disclosure: Anthropic, the safety-focused company behind the Claude family of AI assistants, has said it stopped Claude from being used to help develop biological weapons in different countries.
That is the most concrete public signal yet that AI misuse is not hypothetical, and Gore took it seriously.
He also pushed back on the idea that high-profile warnings from AI leaders are a marketing exercise. "I think we went through a phase where some cynically suspected that the apocalyptic warnings were some kind of bizarre marketing strategy," he said. "If that was ever the case, I don't think it is now." He called Anthropic CEO Dario Amodei "very sincere" and said Sam Altman and Elon Musk were genuine in seconding Amodei's public call to slow the pace of AI capability gains. As AI2Day reported last week, President Trump and Nvidia CEO Jensen Huang dismissed those warnings as a hoax. Gore disagrees, invoking Maya Angelou: "When someone tells you who they are, believe them the first time."
Does he think data center fears are groundless?
Not entirely, but he is precise about where the real concern sits. All AI data centers combined, he said, produce a fraction of the emissions from uncovered landfills worldwide. Air conditioning already consumes more electricity annually than the entire European Union, according to the International Energy Agency, and that demand is expected to triple by 2050.
His sharper criticism lands on hyperscalers, the giant cloud providers building new gas-fired methane turbines to power AI workloads. Locking in new gas plants means decades more of fossil fuel generation. "I much prefer those that are supplying their energy needs with renewables and batteries," he said.
We've followed the tension between data center growth and local communities since August, including the political fights they've sparked nationwide and the global pushback from Scotland to India. Gore's read is that the anger at planning meetings is less about carbon than about anxiety over automation and job losses, amplified by warnings from inside OpenAI and Anthropic.
For anyone trying to follow AI risk beyond the headlines: the biological weapons disclosure from Anthropic is worth reading in full. Political filters, from either direction, are making it harder to assess what's actually in these safety reports, and that's the thing Gore seems most frustrated about.
