Ottawa's online harms bill 'a miss' when it comes to regulating AI chatbots: B.C. premier

B.C. Premier David Eby says the federal government's newly introduced online harms bill "is a miss" when it comes to addressing AI chatbots because it doesn't require tech companies to report users' concerning interactions to police.

"These companies have to have a bright line test imposed on them by the federal government," Eby said in an interview airing Sunday morning on Rosemary Barton Live.

"If someone is using a chatbot to plan a violent crime against others — credibly [and] the company believes that is the case — they have to report, full stop. That should be the law," Eby told host Rosemary Barton.

The B.C. premier noted there are other aspects of the bill his province likes, "but this is a miss" and referenced the horrific mass shooting in Tumbler Ridge, B.C., in February.

After the shooting, it was revealed the shooter's ChatGPT account was flagged for problematic activity before the tragedy, but it wasn't reported to Canadian police by OpenAI.

WATCH | Families of Tumbler Ridge shooting victims file lawsuit against OpenAI:

Families of the mass shooting victims in Tumbler Ridge, B.C., have filed a lawsuit in California against OpenAI. The claim alleges that the creators of ChatGPT failed to notify law enforcement about the shooter’s account activity prior to the attack that killed eight people, including six children.

Bill C-34, also known as the Safe Social Media Act, was introduced Wednesday. It sets out criteria for AI chatbots, including the obligation to reduce the risk of harmful content being generated by their services.

Officials who briefed reporters earlier this week explained that regulated AI chatbots will need measures to respond when a user expresses ideas of suicide, self-harm or an intention to commit an act that could cause death of serious bodily harm to a person.

But the bill does not require the companies behind those chatbots to report interactions to police.

On Wednesday, Culture Minister Marc Miller told reporters the preference is a "very onerous regime" that imposes mandatory transparency to the public about how many incidents platforms are seeing and their reporting requirements.

"OpenAI did have a protocol for reporting out," Miller said. "It just made an egregious human error."

"I'm not going to sit here and pretend today there is one rapid solution that would have prevented Tumbler Ridge from happening, but I do think this law could have made a difference," Miller said.

WATCH | Bill 'could have made a difference' in preventing Tumbler Ridge shooting, says Miller:

Culture Minister Marc Miller said 'I'm not going to sit here and pretend today that there is one rapid solution that would have prevented' the Tumbler Ridge, B.C., shooting, 'but I do think this law could have made a difference.' He said the proposed Digital Safety Act will 'make an onerous requirement on chatbots to interact in a way that is safe.'

Eby told Barton his province will be pushing the federal government on a reporting requirement, and he questioned why the decision to report concerning chatbot interactions should lie with any company representative whose primary relationship is with shareholders.

"I'm not sure why that argument carries any water with anyone," Eby said. "That's why I think a national standard is required."

In the aftermath of Tumbler Ridge, AI Minister Evan Solomon said OpenAI agreed to include Canadian experts in mental health and law within the company's safety office and let experts from the Canadian AI Safety Institute do an assessment of protocols.

The Safe Social Media Act also sets out a plan to ban social media accounts for children under 16 years old unless companies can show they've made their platforms safe. The exemptions will be based on criteria to be determined later through regulations.

In a separate interview, New Brunswick Premier Susan Holt — an advocate of the idea — said "when the research is clear and the tools are in front of us, if we can protect kids from harm, we should be doing everything we can."

Though it will take time for the federal government to establish the criteria and regulator, Holt said there are actions that can be taken now, like disabling social media sites on school internet and trying to make under-16 social media accounts socially unacceptable.

"We're looking at the tools available to us that would be complimentary and supportive," Holt said.

Eby said he is "fully supportive of the proposal that if these companies refuse to put the safeguards in place, they should not have any access to children."

"I think the federal government, on that piece, has struck the right balance," Eby said.

The bill doesn't lay out when the ban will go into effect, nor how the platforms will need to verify users are not younger than 16.

Miller told reporters earlier this week there will "be a back and forth with platforms as to what protects people's privacy and what is adequate and sufficient in the circumstances."