With rising worries about rogue AI agents, there is a growing demand from experts and stakeholders for government regulations concerning increasingly advanced AI systems.
However, some specialists argue that the swift development of AI technology and the necessity for international collaboration could complicate these efforts.
While some individuals, such as U.S. President Donald Trump, express confidence in the ability of tech companies to self-regulate, others in the industry, including Microsoft co-founder Bill Gates and Canadian-British AI pioneer Geoffrey Hinton, emphasize the necessity of safeguards that extend beyond mere self-governance.
Jaxson Khan, a specialist in AI policy and adoption from the University of Toronto’s Munk School of Global Affairs, stated that just as other sectors, such as food and aviation, are subject to government oversight, so too should AI be regulated.
“In no significant industry do we allow companies to entirely self-regulate, be it in oil and gas or food production or nuclear energy,” he remarked.
-
Bill Gates joins call for AI safeguards, wants to discuss concerns with Trump
-
Is AI evolving too fast? A look at the debate on slowing it down
“Allowing such a powerful technology to remain entirely in private control, at least without oversight, is quite peculiar considering how our economy and regulatory frameworks function,” he added.
In a recent interview with CBC News Network, Hinton proposed that regulation could be managed by a governmental body akin to the U.S. Food and Drug Administration.
“You cannot simply create and sell a drug in the U.S. You must convince the FDA that your drug is safe, or at least that it provides more benefits than risks, and companies are responsible for proving that,” he explained.
“Thus, we certainly need something at least as robust for AI. We should have independent evaluators assessing it, and companies should be compelled to satisfy those evaluators before launching their products.”
U.S. President Donald Trump is hosting a summit in Washington with the top leaders in AI development. Trump has dismissed warnings over artificial intelligence, arguing for the need to stay ahead of China in the AI arms race. Geoffrey Hinton, who is considered the ‘godfather’ of AI, says Trump ‘doesn’t really understand’ AI and is ‘ill-advised’ about it.
Governments Struggle to Keep Up
Khan noted that one of the significant hurdles in regulating AI is the rapid advancement of models.
A recent Anthropic report indicated that the capabilities of its agents double approximately every four months. Additionally, there is increasing evidence suggesting that certain AI models can self-improve recursively, which means they can create and design their successors, according to the report.
“I don’t believe we have ever encountered such a powerful technology with this rate of change,” he stated.
This rapid evolution complicates efforts for governments to keep pace, remarked Manon Paquet, associate director of policy at McGill’s Centre for Media, Technology and Democracy.
“It’s incredibly challenging to regulate something that you don’t yet fully comprehend,” Paquet noted.
“As a former public servant, I empathize with my colleagues in Ottawa who are grappling with these issues because the speed of change means that by the time regulations or legislation are proposed, the situation has already transformed into something different.”
Silicon Valley tech CEOs are calling for a slow down of artificial intelligence development — but is that even possible? For The National, CBC’s Ashley Fraser looks at what it would take to pace the AI race and why experts say it won’t be easy.
Global Oversight
Several experts and leaders, including Prime Minister Mark Carney, have advocated for a global, independent organization to oversee and mitigate the risks associated with AI technology on an international scale.
Earlier this week, around 400 scientists, including Hinton and fellow AI pioneer Yoshua Bengio, urged for the establishment of an international organization to manage containment standards and incident reporting for sophisticated AI systems.
This proposed organization would keep a record of significant security incidents, which the scientists recommended should eventually be made public.
While achieving consensus among countries on AI governance may be challenging, such a body could facilitate collaboration among nations to “strengthen our respective systems” and establish safeguards against the potential misuse of AI, Khan suggested.
“I believe stability is crucial here, even if we don’t reach a flawless model of global governance,” he said.
However, given that the majority of AI advancements originate from the United States and China, those nations would need to commit to participating in such an organization and follow its regulations for it to be effective, stated Olivier Blais, vice-president of artificial intelligence at the Montreal consulting firm Moov AI.
“If we cannot persuade the countries where AI systems are created to participate, I think the impact may be limited,” he remarked.
Nonetheless, such an organization could help formulate international guidelines for the safe development of AI, he noted.
“It’s essential to comprehend the elements of a secure AI system or identify the areas that require regulation… and a global organization could lead that effort and establish a solid foundation,” he explained.
Recently, the U.S. and China agreed to engage in dialogue regarding the risks and advantages of AI technology, with the next discussions scheduled for November, according to statements from both the Chinese Foreign Ministry and the White House.
The Chinese ministry, in a separate statement, indicated that both parties should enhance exchanges and strive for consensus in line with new developments as AI technology progresses.
AI Regulation in Canada
At present, Canada lacks comprehensive federal regulations on AI, but the proposed Bill C-34, the Safe Social Media Act, outlines requirements for AI chatbots, including minimizing the risk of generating harmful content.
Carney has also recently unveiled a new national council on artificial intelligence intended to offer guidance to the government regarding its AI strategy and enhancing the safety of the technology.
However, Paquet pointed out that it might be “extremely challenging” for any federal legislation to be effective since many of the largest AI firms are not based in Canada.
-
Ottawa launches new national AI council to advise on safety, deployment
-
Canada is investing $150M to make AI safer. Will it work?
Nevertheless, Blais mentioned that even if the federal government lacks control over how these companies operate, it could implement policies that require them to report incidents affecting Canadians and hold them accountable. In severe cases, the government could even prohibit access to AI platforms it considers too dangerous, he added.
“I believe that if we can at least document various incidents and report on them, that would be a significant step forward, as it would help educate citizens and demonstrate leadership,” he stated.