OTTAWA — Canada is considering signing the first binding international treaty on artificial intelligence. 

The U.S., U.K. and European Union and others all joined the framework convention on AI, human rights, democracy and the rule of law last month. Countries that sign the treaty agree to make rules for how government departments and firms within their borders develop and deploy the technology. 

Canada participated in the negotiation of the convention, which was led by the Council of Europe (CoE), an international human rights organization, starting in 2019. 

Canada is “pleased with the outcomes for this treaty,” said Kevin Sweet, a spokesperson for Global Affairs Canada, adding that it has the potential to set important new legal standards around the use of AI. Sweet said the federal government was “actively assessing” the convention and consulting with provinces, territories and Indigenous groups about whether Canada should sign it.

Policymakers in many of the countries that have signed the treaty have already passed or proposed domestic AI regulations that go further than the convention’s commitments. The convention is meant to act as a backstop as governments update their laws in response to the potentially disruptive technology. It notes that AI has the “potential to promote human prosperity” and innovation, but carries significant risks.

Signatories commit to requiring that AI systems are reliable, don’t discriminate and respect privacy. They also agree to impose transparency measures such as ensuring any machine-generated content is properly labelled and disclosing to people when they’re interacting with AI. The treaty includes a carve-out for governments using AI in applications related to national security.  

The Liberal government’s proposed Artificial Intelligence and Data Act includes similar requirements for “high-impact” settings like healthcare, employment and policing. But it doesn’t outright ban any uses of the technology. The convention requires countries to look at whether there are some cases in which using AI systems would be so harmful that it shouldn’t be allowed.

The Yukon government backs Canada signing the treaty, which “strikes a responsible balance between the benefits and risks of AI,” said spokesperson Francis Lefebvre. Quebec and Nova Scotia are analyzing the convention, officials said. British Columbia, New Brunswick and Saskatchewan declined to comment on whether they support the initiative, citing rules around recent and upcoming elections. 

While the CoE claims the convention is the first binding AI treaty, it’s one of several international efforts to set ground rules for the technology. The United Nations, OECD and G7 have published high-level principles for the development and use of AI. Ottawa has sought a prominent role in international policy conversations about AI, and is set to launch a new safety institute to study the technology.