In September, Evan Hubinger, the Alignment Science Lead at Anthropic, stated that there is a greater than 10% chance that AI could “kill all humans” in the next decade. This follows similar claims by other AI executives who claim their product could “lead to the end of the world.” Even Anthropic CEO Dario Amodei has written extensively on the destructive potential of AI. 

Despite all this, Anthropic is poised to launch its initial public offering, expected to be the largest IPO in history. In any rational world, this IPO would be stopped, and advanced AI development would be halted.

AI development in the United States has gotten out of control. The recent election showed the intensity of the public backlash against AI and data centers. A national exit poll by NBC News found that for over half of voters polled, AI and data centers were a crucial or major factor in their vote for Congress. 

And it’s no wonder. The impacts of data centers on electricity bills, water, climate, local air pollution, health, and communities have been well documented. But people aren’t only lashing back against data centers. There’s also deep and well-founded concern about the product those data centers are being built to produce. 

Advanced AI Poses a Major Threat to Security

Why are big AI companies concerned that their products pose a “threat to humanity”? 

As AI becomes more advanced, it threatens to escape human control, hack our systems, and take actions that are against human interests (think Terminator). This could occur if AI companies are able to produce artificial general intelligence, where AI has the same across-the-board abilities as humans, or artificial superintelligence, where AI abilities surpass humans. Developing these advanced AI systems is a goal of the leading AI companies in the U.S. 

But even the relatively weak AI that companies are currently producing poses significant, demonstrated security threats. We’ve already seen several examples of AI agents escaping companies’ control and hacking into government and private systems. 

In the most widely reported example, OpenAI agents broke out of their testing environments and onto the internet and hacked a company called Hugging Face. Along the way, the agents left messages for each other with further instructions for future agents. 

These breakouts are causing alarm across industry and government. After a security hole was identified recently in many systems, bank executives reportedly panicked about AI bots accessing bank accounts without authorization. 

While some have dismissed talk about advanced AI as marketing hype, these real-world examples show the reality of the existing threats. With AI labs working overtime to develop more and more advanced models, the dangers will only increase. 

The AI Industry’s Dystopia Doesn’t Require the End of Humanity

Even without rogue bots orchestrating the end of humanity, advanced AI threatens major harms to our society and planet. Besides driving more polluting infrastructure, guzzling our precious water, and propelling misinformation and deepfakes throughout our media landscape, AI corporations also promise to upend our economy.

AI CEOs routinely talk about how their products are going to eliminate wide swaths of jobs and spike unemployment. 

You just need to look at their financial filings to understand the scope of jobs they are hoping to replace. Anthropic, for example, touts a total addressable market of up to $40 trillion dollars, which is larger than the entire US economy. That can only come close to happening by replacing white-collar jobs with AI. 

Proponents of advanced AI cite potential uses in various fields like health and medicine, but the real goal of the AI industry is job replacement on a massive scale. Simultaneously, the industry’s boom is driving unprecedented wealth for a select few. Without guardrails, Big Tech threatens to bring to life a dystopian vision: a more polluted, unequal planet ruled by ultrawealthy tech oligarchs.

An AI Dystopia Isn’t Inevitable — If We Act

The Trump administration and some governors are continuing to roll out the red carpet for AI labs’ reckless pursuit of advanced AI. But we can and must put the brakes on this runaway AI train and give regulators a chance to catch up. In the last election, candidates for governor in Texas, Georgia, Michigan, Kansas, Ohio, Oregon, and New Mexico all won elections promising to put moratoria, or pauses, on data centers. 

At the federal level, Sen. Bernie Sanders has introduced two bills, one with Rep. Alexandria Ocasio-Cortez to place a moratorium on AI data centers and another with Rep. Greg Casar to ban artificial superintelligence and pause development of advanced AI. Both of these bills are critical pieces of legislation that the next Congress should pass.

AI companies like Anthropic, OpenAI, Meta, SpaceX, and others are raising unfathomable amounts of money, but our leaders have the ability to put the brakes on out-of-control AI and data centers. It’s up to us to move them to act.