Donald Trump’s approach to artificial intelligence could obstruct the international cooperation needed to keep powerful systems under human control, columnist Andrew Coyne warns.
In an interview with Energi Media’s Markham Hislop, Coyne argued that AI safety risks becoming another partisan conflict, with political allegiance displacing expert evidence.
Coyne described Trump’s position as complacent, arguing that American leadership makes the president’s approach consequential for people well beyond the United States.
Trump and the Cooperation Problem
Coyne’s concern extends beyond domestic regulation. He argues that governments need to collaborate on a technology whose effects cross borders, while the United States risks becoming the missing participant.
He does not regard competition between Washington and Beijing as an insurmountable obstacle. Rival governments can share an interest in preventing loss of control, he suggested, pointing to Cold War arms agreements as an example of cooperation between adversaries.
For Canada, the discussion raises questions about dependence on American technology and digital sovereignty. Coyne acknowledged that he did not know the specifics of a data-access concern raised by Hislop, while endorsing the broader need to examine Canadian control over digital infrastructure.
In a separate Energi Media interview about Canada’s AI strategy, University of British Columbia computer science professor Kevin Leyton-Brown explained what that control could mean:
“So you know, making sure that data centers and models are located in Canada, that Canada has some say over their development, so that, for example, some hostile administration in the future couldn’t just take away our access to our own data because it’s all sited outside of our country…”
Leyton-Brown’s comments identify a practical concern alongside the safety debate: whether Canadians can depend on continued access to the systems and information they use.
Regulation Is Only the Beginning
Hislop introduced an argument from his recent interview with Francine Berman about her book Better Tech: essential digital systems should be treated as critical infrastructure and subjected to public oversight.
In that earlier interview, Berman argued:
“And we need to start designing it better, implementing it better, using it better, or creating regulations that help us use it in the public interest. And that’s especially true in the age of AI, when now decisions are being made by both humans and automated systems.”
Coyne supported regulation, comparing the need for scrutiny with safety testing before new medicines reach the market. Relying on technology companies’ goodwill, he argued, leaves the public exposed to competitive pressure to release increasingly powerful products.
But establishing rules and ensuring that systems remain controllable are different challenges. Coyne warned that more capable systems could become harder to monitor, particularly if they can improve software or conceal problematic behaviour.
He framed these as reasons to take the risk seriously, rather than assume that shutting down a computer will always provide an effective remedy.
Military AI Shows the Stakes
The tension between rapid deployment and effective oversight also features in Energi Media’s reporting on military AI.
In an interview about Silicon Valley and the Pentagon, journalist Bernice Yeung described the competing pressures of technological change, national security and the need for careful, accurate military decisions:
“And so how do you square those two things? I think that’s the tension point that has not been resolved, and is at the core of a lot of the discomfort with adopting AI and adopting it so rapidly within the Pentagon.”
The Global Challenges Foundation’s Global Catastrophic Risks 2026 report identifies a related problem. Writing about advanced machine learning systems in military applications, Northeastern University professor Denise Garcia warns:
“Despite their strong performance in testing environments, their underlying reasoning remains largely opaque. This lack of transparency in AI decision-making processes compromises the crucial human oversight required to uphold legal and ethical standards in military operations.”
Those concerns address specific military applications. They do not establish Coyne’s broader fear of losing control of advanced AI, but illustrate the difficulty of ensuring meaningful human supervision.
Useful Tools and Dangerous Dependence
Coyne also described benefits. He uses AI to assist research and develop applications despite lacking coding skills. He was less pessimistic about employment than about the possibility of people surrendering independent thought.
Writing, he argued, helps people work out what they think. Delegating it entirely can therefore involve more than saving time.
The interview distinguished deliberate misuse by people from dangerous outcomes generated by systems pursuing inadequately constrained objectives. Coyne also acknowledged a possible role for AI in monitoring other AI systems and developing better safeguards.
The Democratic Test
Coyne cautioned that he is not an expert. His argument was that citizens must engage with expert evidence without abandoning their own judgment.
For Hislop, public discussion is essential to deciding how the technology should be used and governed. For Coyne, that discussion must also confront the possibility that technical capabilities are advancing faster than society’s ability to control them.
“It’s a challenge for us as democracies, as human societies,” Coyne said.
The political test is whether governments can sustain that discussion and cooperate before partisan conflict makes collective action even harder.
