By Maria Cheng
OTTAWA, June 12 (Reuters) – Canadian legislation to regulate AI chatbots, introduced this week after national outrage over a school shooting that left nine people dead, has raised doubts among academics and legal experts who warn that loopholes and its lengthy time frame may undermine the effort.
Public pressure mounted on the government for action after OpenAI acknowledged it had not reported to police troubling ChatGPT messages from the suspect in the February incident. The bill proposes a new digital regulator and would follow Australia in banning social media for children under 16.
The new agency would require chatbots to reduce the risk of users seeking harmful content and include crisis intervention steps if they discuss issues such as suicide and self-harm.
But the legislation has drawn criticism for its lack of detail, the difficulties of avoiding loopholes while preserving privacy, and the long time frame for putting measures in place.
“If this is the preview of a law, I do not have high hopes for something that will be useful in a practical sense,” said Evan Light, an associate professor at the University of Toronto who focuses on technology and privacy.
He said he was shocked at how underdeveloped the bill was, adding that restrictions on Internet use could easily be circumvented with VPNs – virtual private networks that mask users’ Internet addresses – or other means.
The office of Canada’s Identity and Culture Minister Marc Miller did not immediately respond to a request for comment on the issue.
In a news conference announcing the bill on Wednesday, Miller said there was a need to balance privacy concerns with efforts to regulate social media and AI chatbot use, and noted the proposed law does not apply to private messaging apps like WhatsApp or Signal. He also said companies that meet criteria set by the new regulator would be able to get an exemption from the social media ban.
He said Canada had witnessed the tragic consequences of online harm, citing the shooting at Tumbler Ridge in British Columbia. The 18-year-old suspect’s account on OpenAI’s ChatGPT had been flagged for violence internally, but was never reported to police.
Miller said the company “made an egregious human error” and the government was now taking steps to make social media and AI chatbots “safer by design.”
OpenAI did not respond to a request for comment. The company previously apologized for not alerting Canadian authorities.
A growing list of governments across Europe, in Brazil and in a handful of U.S. states are pushing ahead with new age-checking requirements for social networks, AI chatbots and porn purveyors alike.
A Meta spokesperson said in an email that social media bans are “counterproductive” and that the company is assessing the implications of the proposed bill.
A spokesperson for Google, which owns YouTube, said the company is committed to working with the government to establish higher safety standards, so parents can choose safer online experiences for their children. A TikTok spokesperson said the platform already has more than 50 safety and privacy settings and parents can use “Family Pairing” to manage their teens’ experience.
Government officials have said it could take a year for the bill to pass and 18 months to set up the digital regulator once it does.
Florian Martin-Bariteau, director of the Centre for Law, Technology and Society at the University of Ottawa, warned that children would likely be able to circumvent the social media ban and AI chatbot restrictions.
Since Australia’s social media ban took effect, the regulatory agency enforcing it reported that a substantial number of children under 16 still retain accounts.
“The proposed framework will move them to riskier, smaller platforms,” he said, adding that while rules for major platforms could likely be enforced, it would be virtually impossible for Canada to block smaller websites that don’t comply. “By trying to protect kids, we may actually put them at greater risk.”
(Reporting by Maria Cheng; Editing by Caroline Stauffer and Edmund Klamann)
Comments