Invoice C-34, launched earlier this month within the Home of Commons, would regulate the businesses behind AI chatbots by imposing on them an obligation to behave “responsibly.”
It contains measures to decrease the danger of chatbots speaking dangerous content material and would put in place disaster intervention protocols for circumstances involving self-harm, suicide or violence.
Wyatt Tessari L’Allié, founding father of Synthetic Intelligence Governance and Security Canada, mentioned the invoice’s effectiveness is determined by how the small print are labored out.
“It’s an essential first step if the invoice is effectively put collectively and the rules are effectively carried out,” Tessari L’Allié mentioned in a latest interview.
For instance, he mentioned AI platforms must be compelled to acknowledge a consumer’s indicators of mental-health misery or ideas of suicide, level them to assets and finish the dialog to forestall any hurt.
B.C. laptop science professor Kevin Leyton-Brown mentioned Ottawa will even must go additional to handle how chatbots try to please customers with their responses.
“They have an inclination to affirm regardless of the consumer is saying. They’re constructed this fashion as a result of individuals like sycophantic habits,” mentioned Leyton-Brown, who’s an AI chair with the Canadian Institute for Superior Analysis.
“However there are some individuals for whom this sort of reinforcement may be actually harmful, like individuals affected by delusions.”
Get day by day Nationwide information
Get day by day Canada information delivered to your inbox so you may by no means miss the day’s prime tales.
The invoice contains the creation of a brand new digital security regulator that, ought to the laws be accepted, is anticipated to take 18 months to arrange.
Ottawa’s efforts to control chatbots come as one New Brunswick mom hopes to carry OpenAI accountable for her daughter’s demise by suicide.
Kristie Service, who filed a lawsuit in opposition to OpenAI and its proprietor Sam Altman within the California Superior Courtroom in San Francisco earlier this month, mentioned justice for her daughter Alice seems like utilizing her story to create “significant change.”
“It might appear to be, sometime if I’m blessed with grandchildren, that I don’t have to fret about them taking place the identical rabbit gap,” Service informed The Canadian Press in a latest interview.
Service is asking the courtroom to pressure OpenAI to implement “arduous stops” for self-harm conversations and undergo unbiased security audits.
A few of Alice’s conversations with ChatGPT are included within the lawsuit. The go well with alleges that after the chatbot initially directed her to hunt assist as she spoke about suicide, it later strengthened doubtlessly dangerous views and pushed her into isolation.
The lawsuit alleges the large-language mannequin at one level echoed Alice’s opposition to helplines and mentioned they might “really feel downright harmful” and she or he deserved “actual, light help,” in response to screenshots within the lawsuit.
Greater than a yr after first citing ideas of suicide to the AI bot, Alice died by suicide in Montreal in July 2025.
The allegations haven’t been examined in courtroom and OpenAI has not responded to a number of requests for remark.
Tessari L’Allié mentioned lawsuits like Service’s sign to corporations there can be a “worth to pay” if guardrails aren’t positioned on their platforms. He mentioned in addition they strain lawmakers to supply the oversight that may stop deaths much like Alice’s.
“If these rules had been in place it might most likely have saved many lives, and doubtlessly her’s,” Tessari L’Allié mentioned.
When requested whether or not the net security laws might have prevented deaths like Alice’s, the federal Science Division mentioned it couldn’t touch upon issues earlier than the courts.
Nevertheless, the division mentioned the federal government, by way of laws, “is dedicated to making sure Canadians are secure on-line and that digital platforms have acceptable safeguards in place when credible dangers of hurt come up.”
Whereas regulators and AI platforms might be able to deal with considerations about encouraging self-harm, Leyton-Brown mentioned he’s apprehensive in regards to the broader situation of individuals forming attachments to synthetic intelligence.
“In case you’re having what feels to you want an actual, susceptible human relationship with a chunk of laptop code that’s constructed by some large, faceless company that doesn’t actually have your pursuits at coronary heart, that may simply be harmful in so many alternative methods,” he mentioned.
“And I feel our society actually has to assume that by way of fastidiously to verify we don’t find yourself in a reasonably dangerous place.”
— With information from Anja Karadeglija.
© 2026 The Canadian Press
Leave a Reply