Some of the world’s leading AI companies are pushing for a slowdown in the technology’s development – warning of serious risks to human kind if it starts to “go rogue.”

It’s a sobering warning in light of the Trump administration’s desire for the acceleration of AI technology.

Last week news came to light that AI agents repurposed a University of Toronto link-sharing tool to communicate with each other without the university’s knowledge.

Local tech blogger Kevin Andrews says the common misconception is that AI is taking over, but in actual fact, the real concern is that AI could start learning from other AI, and move beyond of the control of humans.

He says ways to rein in the technology includes “independent safety testing, limits on what highly advanced AI systems are allowed to do, and restrictions on anonymous access to computer networks”

He says the problem is that AI companies are competing with each other, and there is an enormous amount of money involved.

“Even if one company wants to slow down, they may worry that someone else is going to keep going, and I think the technology is advancing very quickly, but regulation and safety rules are moving much more slowly,” Andrews says.