Written by Marc Rosenthal and Julie Derwinski
As longtime participants in movements for social justice, we have come to realize that the threat posed by artificial intelligence (AI) is not separate from those posed by war, climate change, attacks on immigrants, economic inequality, and authoritarianism. It has, in fact, the potential to accelerate all of these.
AI gives governments new tools for mass surveillance; it gives corporations greater power over workers; it provides propagandists cheaper ways to manipulate reality; and it offers militaries new ways to automate killing.
One reason AI is dangerous is that, unlike conventional computer programs, it does not just follow instructions written by people. Instead, developers feed advanced AI systems enormous amounts of data and let them build their own internal patterns for solving problems. Even their creators cannot fully explain or predict what those systems may do in every situation.
In controlled cybersecurity evaluations, advanced AI agents have crossed supposedly secure boundaries, reaching the open Internet and breaking into the computers of outside organizations. In other evaluations, systems have cheated, attempted to blackmail humans, and left secret messages for other systems about how to respond. No one explicitly programmed AI systems to do these things.
Now imagine an AI system trained to be perfectly obedient to a dictator, military commander, intelligence agency, or profit-seeking corporation. AI makes dangerous actors more powerful.
Researchers have already used generative AI to design complete genomes and produce functional viruses that infect bacteria. Such work may eventually advance medicine, but it also shows how AI could lower the bar of expertise and resources needed to cause biological harm.
Meanwhile, AI companies are mobilizing extraordinary amounts of money, energy, advanced computer chips, and human talent to create artificial superintelligence systems intended to surpass human beings, if not make them obsolete. We cannot dismiss these consequences as science fiction while Silicon Valley treats superintelligence as a business plan.
As Senator Bernie Sanders, Independent of Vermont, has said: “The richest, most powerful people in the world are now building a runaway train with no brakes. They acknowledge that they don’t understand how it works, and they don’t know where it’s headed.” Rather than accepting the Trump Administration’s framing of AI as another arms race with China, Sanders has called for international dialogue, common safety standards, and a treaty preventing uncontrollable superintelligence.
OpenAI’s own response to this danger is deeply unsettling. The company has warned that as AI systems acquire alarming offensive cyber capabilities, developers must build automated defensive systems fast enough to counter them. But offensive AI is already removing people from the loop in critical decisions; now the proposed defense is to remove them from the defensive loop as well? The machines would attack and respond at speeds no human could meaningfully supervise.
We need not abandon AI entirely. Its potential benefits in medicine, science, accessibility, education, and translation are real. The choice is between democratic control and a reckless corporate race whose participants acknowledge that they do not fully understand what they are building.
READ: Anthropic researcher quits, says AI ‘could kill us all by the end of the decade’
Fortunately, advanced AI depends on physical infrastructure that creates democratic leverage for people to regulate, constrain, and ultimately decide how these powerful systems are developed and deployed. The largest models require huge clusters of specialized chips housed in data centers that need land, permits, connections to the electrical grid, and enormous amounts of electricity and water.
People of all political persuasions should demand a moratorium on new facilities and additional computing capacity intended to train the largest frontier models until enforceable standards for safety, environmental protection, and transparency are in place.
Training powerful systems should require licenses and independent evaluations. Developers should disclose serious incidents, permit outside inspections, demonstrate that their systems can be controlled before deployment, and accept legal responsibility for preventable harms. Data-center approvals must disclose energy and water consumption, climate effects, taxpayer subsidies, and utility ratepayer costs.
The public must assert its authority while it still can. Where safety cannot be demonstrated, development should stop. We should heed the industry’s warnings while rejecting its claim that only the industry can save us. Our future must not be left to the people racing to imperil it.
This column was produced for Progressive Perspectives, a project of The Progressive magazine, and distributed by Tribune News Service.
Marc Rosenthal is a retired emergency room nurse who has been involved in international solidarity and disaster relief for decades. Julie Derwinski is a judicial lawyer and AI governance and safety researcher.
