Artificial intelligence: the Quebec electoral officer calls for better legislative oversight

People arrive at a polling station in Montreal, Sunday, Oct. 26, 2025, as advanced voting begins ahead of the upcoming municipal election on Nov. 2. (Graham Hughes/The Canadian Press)

Disinformation, deepfakes, foreign interference: Quebec’s Chief Electoral Officer (DGEQ), Jean-François Blanchet, says he takes artificial intelligence (AI) and its potential impact on future elections seriously.

At the same time, he admits that his resources are limited and calls for better oversight from elected officials.

In a context where the spread of false information can be accelerated by AI, Blanchet said that voters have a responsibility to obtain information from reliable sources.

At the same time, he is keen to be reassuring.

“Once voters have made their choice and cast their ballot, the result will reflect the will of the voters,” he said in an interview with The Canadian Press.

The problem lies more in what happens before voters enter the polling booth, as this is where misinformation—whether fuelled by AI or not—can do its work.

“This is the part that is difficult to control. (...) We can’t control what spreads on social media,” he said.

Blanchet said that his organization is working to improve its content so that “there are no grey areas.”

“We will ensure that, for everything related to the electoral process, the correct information is provided,” he said.

A more robust legislative framework?

Bill 98, passed in May 2025, created an offence for anyone who knowingly disseminates false information with the intent to influence an election or undermine public confidence in the electoral process.

However, the DGEQ concedes that bringing charges remains difficult.

“In criminal cases, we must be able to prove beyond a doubt that an offence was committed intentionally, and find out who did it. So it will always remain difficult,” he explained.

Blanchet believes, however, in the need for a legislative framework for AI, despite the difficulties.

“Why hasn’t it been done until now? I think legislators don’t know where to start,” he argues.

The DGEQ added that political parties will have to “take responsibility” because they have a “major role to play in protecting democracy.”

“In cases where false information about the party is circulating, it will also be up to them to set the record straight,” said Blanchet.

Élections Québec will use AI

But we must not “bury our heads in the sand” and recognize that AI will be increasingly used by everyone, stresses the DGEQ.

And Élections Québec will be no exception.

The October 2026 elections will be the first in which the public body will make use of this technology.

Blanchet explained that Elections Québec will hire 70 people whose job will be to answer voters’ questions, and AI will help them in their work. The DEGQ is reassuring: the use of AI will be limited and supervised. Employees will be able to ask questions to a chatbot that will only have access to internal information from Elections Québec.

Will voters one day be able to ask questions directly to AI to obtain information about elections?

“Inevitably, I think we will get there one day,” said Blanchet.

But he assures us that there will always be humans behind this AI.

In the meantime, the DGEQ warns voters against using conversational agents such as Copilot, Gemini or ChatGPT. He says that Élections Québec received complaints from citizens who obtained incorrect information from AI during the last municipal elections.

Concrete risks

The risks of AI on elections are very real. The Commission on Ethics in Science and Technology (CEST) documented some of these risks in a report published in 2024.

It points out that “since generative AI (such as ChatGPT) can easily create content that can deceive users, it greatly exacerbates the risks of political manipulation through disinformation.”

“Disinformation in digital media fuelled by AI is a form of manipulation, as it aims to insidiously alter beliefs by exploiting asymmetries in knowledge about how AI and digital media work. It thus undermines voters’ ability to make autonomous choices and participate in an informed manner in public deliberation,” the report states.

The CEST notes, for example, that generative AI tools can be used to intimidate and thus discourage citizens from entering politics.

“A large proportion of deepfakes are in fact non-consensual deepfakes, often of a pornographic nature, and in the vast majority of cases they target women. This type of online hostility towards women can cause some to question their participation in public life,” the brief adds.

The CEST’s mandate is to advise the Quebec government on all matters relating to ethical issues in science and technology.

This report by The Canadian Press was first published in French on March 8, 2026.