Need for Safe and Productive Development and Use of Artificial Intelligence

Honourable senators, I would like to begin by thanking Senator Moodie for initiating this inquiry and for calling attention to the need for the safe and productive development and use of artificial intelligence in Canada.

Artificial intelligence, or AI, has become ubiquitous in society. Every day, we consume information generated by AI, or we may talk with an AI bot. AI is summarizing our health appointments for doctors and pulling research for students. It is also reportedly firing missiles in Iran.

On June 8, Forbes released an article entitled “Tech Rivals Unite To Stop AI-Designed Bioweapons.” The article explains that the executives of major tech companies, such as OpenAI, Anthropic, Google DeepMind, Microsoft AI and Meta, all signed an open letter to Congress. In summation, Forbes writes, “. . . the technical guardrails keeping biological weapons out of the wrong hands are rapidly eroding.” Indeed, Senator Moodie, this inquiry is timely, relevant and important.

I wanted to wait a bit before participating in this inquiry. By waiting, I was able to participate in important discussions about the intersection of AI and First Nations perspectives and attend a discussion about AI and safeguarding democracy. I benefited from the newly released Canadian AI strategy: AI for All. And I benefited from Senator Hay’s excellent learning seminars entitled “AI Explained.”

Last week, I attended an event at Mila, the Quebec Artificial Intelligence Institute and the largest AI research facility in the world. After the meeting, I had a fascinating conversation on the concept of agentic AI with Dr. Hause Lin, a research scientist with MIT and Cornell University. He is also a fellow with Mila.

This type of AI, as it was described to me, pursues a goal and makes its own decisions along the way, much like an assistant or agent you hand a task to and let run with it. These AI agents already exist and are quite popular, but they come with risks.

Iason Gabriel and Arianna Manzini released a paper in 2024 entitled “The ethics of advanced AI assistants,” which posits human control must be maintained to ensure that our morals and ethics remain at the heart of decision making.

The paper asserts:

AI assistants will likely have a significant level of autonomy for planning and performing sequences of tasks across a range of domains. Because of this, AI assistants present novel challenges around safety, alignment and misuse.

With more autonomy comes greater risk of accidents caused by unclear or misinterpreted instructions, and greater risk of assistants taking actions that are misaligned with the user’s values and interests.

More autonomous AI assistants may also enable high-impact forms of misuse, like spreading misinformation or engaging in cyber attacks. To address these potential risks, we argue that limits must be set on this technology, and that the values of advanced AI assistants must better align to human values and be compatible with wider societal ideals and standards.

Pillar 1 of the new AI strategy is called Protecting Canadians and Safeguarding Democracy.

To accomplish this, I believe that any legislation must take a forward-thinking, future-proofing approach to this rapidly shifting technology. We cannot pass legislation that approaches AI as though it is static, rather than evolving.

Two years ago, I could not have imagined the way AI would be interwoven into everything we do. Similarly, two years from now, AI will no longer be the same as we know it to be today.

Agentic AI is quickly becoming the way of the future, whether Canada is ready for it or not. We must also turn our minds to legislation that contemplates the use of “bossware,” surveillance pricing and dark patterns, issues that many of us are only starting to learn about.

Safeguarding democracy will require legislative and policy reform to counteract mis- and disinformation campaigns through the use of AI “deepfakes” and AI bots to manipulate voter views. It will require us to invest more into research and ensure that our researchers have access to key data currently being held by large entities such as social media companies.

Under the European Union’s Digital Services Act, very large online platforms and search engines are legally required to share platform data with qualified, vetted researchers. However, according to some researchers, very little data, if any, has been shared. It will take political will to ensure that researchers have the resources and data required to study systemic risks, like election manipulation, illegal content and the spread of misinformation.

Pillar 2 speaks to “Empowering Canadians” through education initiatives and targeted investments, while Pillar 6 speaks to “Building trusted partnerships and global alliances.”

Importantly, in October 2025, a joint letter was signed by “. . . more than 160 academics, experts, advocates, civil society and human rights organizations across Canada . . .” who felt the 30‑day consultation period was woefully inadequate. Fenwick McKelvey, an associate professor of information and communication technology policy at Concordia University, pointed me toward the site of an initiative called People’s Consultation on AI, self-described as “. . . a parallel and independent process . . .”

It seems, from those submissions, that trust in the government’s approach was lacking before the strategy was even released. If you do not start from a place of trust, it will be difficult to build trusted partnerships.

After publication of the strategy, a June 6 article in The Walrus was published with the title “Canada Finally Has a National AI Strategy. Experts Hate It.” Nearly every contributor decries the government’s approach to forge boldly ahead with AI adoption while belatedly engaging the public through education initiatives on the assumption that trust will naturally follow.

As Alissa Centivany from the Starling Centre for Just Technologies and Just Societies says in the article, “The strategy misconceives public trust as a means to an end.”

The article continues:

A recent scan we undertook showed over 218 AI applications being trialled and used by the Canada Revenue Agency alone. Many of these are licensed from private companies, whose business interests are opaque even to the civil servants who implement them. . . .

While I do not think Canada can afford to fall behind other countries that are quickly adopting new technology to streamline their inner workings, we cannot scale too much too quickly.

The article includes a cautionary story of a bot used by the CRA that was later pulled, as it was found to be incorrect 67% of the time, reinforcing that many around the world are now putting their trust in still-developing systems.

I heard a story of a person who took a photo of a fresh cut on their finger and asked AI what to do. According to Dr. Lin in an email exchange:

Because agentic AI can think, decide, remember and even offer comfort, it’s easy to let it; but every ability we stop using weakens, like an unused muscle.

Pillar 4, “Building the Canadian sovereign AI foundation,” and Pillar 5, “Scaling Canadian champions,” are both aimed at improving sovereignty over AI.

To ensure sovereignty, we must be mindful of every action. Disturbingly, the Toronto Star published a series of investigative reports on the tens of millions of dollars that DND has spent with the U.S. data and AI company Palantir, the same platform the U.S. military uses to power and coordinate weapons.

The article in The Walrus also says:

. . . it is difficult to see how the government can remedy the harms of consolidated tech power and wealth anytime soon.

This raises an often-repeated question for those critical of the wholesale embrace of AI: What policies, legislation, regulations and investments must we make to ensure that the benefits of AI are not concentrated in an elite few? How can we ensure that Pillar 3, “Powering AI adoption for shared prosperity,” is achieved?

A report published by the University of Oxford entitled Agentic Inequality speaks to the potential for agentic AI to foster inequalities in power, opportunity and outcome. Computing power and the associated infrastructure all cost money. AI companies have rapidly accumulated wealth and are valued at numbers so high, they have some questioning whether the AI bubble will soon burst.

The abstract of this report explains:

This paper introduces and explores “agentic inequality” – the potential disparities in power, opportunity, and outcomes stemming from differential access to, and capabilities of, AI agents. . . .

Already, this concept has been raised by First Nations, who point out that “AI for All” is not achievable for those communities that cannot overcome the infrastructure divide. Without targeted investments and co-developed strategies around data governance, data sovereignty and compute ability, First Nations can actively be left behind in the new AI age, adding to the already sharp disparities between First Nations and non‑Indigenous communities.

But it’s not all doom and gloom. A 2024 paper by Michael Henry Tessler et al., entitled “AI can help humans find common ground in democratic deliberation” and published in Science, describes research demonstrating how AI systems can help diverse groups find and reach consensus.

I believe that technology itself is neutral. However, development without guardrails is dangerous. Well-funded research will be essential.

These conversations, events and readings have led me to one conclusion: There is a lot more I would need to know before feeling confident when voting on any legislation involving AI. There are many ethical, social and other far-reaching implications with this technology; I am hopeful that we can get the guardrails we need in place quickly enough.

Thank you, wela’lin.

(On motion of Senator Kingston, debate adjourned.)