Synopsis
Canada's proposed law requiring chatbot providers to protect users from "harmful" content is the latest step by global governments to regulate the sector. There are also many state-level bills, including one passed in New York that would require protections for minors against harmful content as well as so-called "sycophancy" and age verification requirements for chatbots.Listen to this article in summarized format
The draft legislation introduced on Wednesday targets issues like chatbots encouraging suicide or self-harm, as well as AIs deceptively claiming to be human or expert in complex fields such as medicine and law.
Chatbots will also be subject to rules from a new Digital Safety Commission.
"I think it's the first online harms regime to add chatbots to the scope," said Taylor Owen, Beaverbrook Chair in Media, Ethics and Communications at Montreal's McGill University.
"We're pretty late to this policy space in Canada, so the core of the bill is able to learn from what a lot of other countries have tried," he added.
Here is how Canada's proposed rules stack up against other jurisdictions:
Britain
The British government said in February that it would "shut a legal loophole" leaving chatbots uncovered by the requirements of last year's Online Safety Act.
That legislation includes requirements for age checks on content such as pornography or promotion of self-harm, suicide or eating disorders, and bars on children seeing bullying or hateful messages.
London is also looking for flexibility to introduce new rules without changing the law every time technology changes -- similar to the guidelines to be set by Canada's new independent commission.
"You have to be really careful not to overdraft in legislation, or you have to go back to parliament to change anything, and that can take years," McGill's Owen said.
United States
Federal lawmakers in Washington have introduced multiple bills to regulate chatbots, including the GUARD and CHATBOT acts in the Senate and the SAFEBOTS act in the House.
With Congress divided and tech firms lobbying heavily against regulation, such bills face an uphill battle to become law.
While exact provisions vary, most would forbid sexualised interactions, self-harm or suicide content for minors, with some requiring parental consent or shared "family" accounts.
There are also many state-level bills, including one passed in New York that would require protections for minors against harmful content as well as so-called "sycophancy" and age verification requirements for chatbots.
Australia
After Australia became the first country to ban social media for under-16s, some politicians are now pushing for the youth curbs to be extended to chatbots.
"Now is the chance to get ahead of the curve on AI and not be mopping up the damage in 15 years' time," independent MP Kate Chaney told Australia's 9News.
EU
Although its comprehensive AI Act covers chatbots, the EU went further this year after widespread outrage at xAI's Grok chatbot creating sexualised deepfake images.
The bloc has approved rules to ban AI systems generating sexualised deepfakes, and there is currently an EU probe into Grok over the scandal.
The European parliament has warned of "significant risks to minors' rights and healthy development" from AI, and there are calls to better protect children online.
If Canada's law goes through, "the nature of these regulations... is that countries learn from each other and then iterate," Owen said.
"We will see what works and doesn't here, and then other countries can learn from that, just like we did."
Among individual EU member states, Italy has barred companion chatbot provider Replika over data protection and age restriction concerns.
Brussels has in recent months spent a lot of political energy across the EU on debating whether to set minimum ages for unfettered access to social networks.
EU chief Ursula von der Leyen said in May the bloc should explore limiting children's access to social media with possible new rules proposed within months, speaking before recommendations from an experts' panel expected to be published in July.
China
Adding to its tight control of online spaces, Beijing has been among the fastest to introduce restrictions on minors' access to AI chatbots.
New rules applicable from July 15 bar chatbot operators from offering virtual companions or romantic partners to under-18s, state broadcaster CGTN reported in April.
The Cyberspace Administration of China will also require consent from parents or guardians for any human-like chatbot interaction with under-14s.
And the government will scrutinise algorithms powering chatbots, demanding operators reveal them to officials.
