Meta to alert parents if their child discusses suicide with AI chatbot

Meta Platforms Inc. has launched new tools to alert Canadian parents and first responders when youths are at risk of suicide or self-harm, but experts say they’re not infallible.

The new measures, which were rolled out Thursday in Canada, the U.S., U.K. and Australia, notify parents using Instagram supervision tools if their teen is discussing suicide or self-harm with chatbot Meta AI.

The tech giant is also building a way to contact emergency services when conversations adults and teens are having with Meta AI suggests they may be at risk of taking their own life.

The features come after several high-profile instances of children researching self-harm and, in some cases, being coaxed into suicide through artificial intelligence-based chatbots. 

While experts applauded Meta for taking steps to protect kids ahead of Canadian legislation establishing safety requirements for social media and AI chatbots winding its way through parliament, they say the effectiveness of Meta’s measures might be limited.

That’s because not every parent uses the Instagram supervision tools that make suicide and self-harm alerts possible and because kids are prone to finding workarounds.

Because Meta isn’t offering the protections by default, it comes off as “care washing,” said Kaitlynn Mendes, a sociology professor at Western University and the Canada Research Chair in inequality and gender.

“It’s like look at all the great things we’re doing for these accounts which might not be used in practice,” she said.

Even when the accounts are used and self-harm risk can be flagged, she suspected some users will be clever about escaping detection. For example, they may say it’s a fictional character contemplating self-harm rather than themselves.

“If people are really that interested in finding ways to engage in self-harm or mass violence, they’re probably going to be persistent,” she reasoned.

Shauna Pomerantz, a professor of child and youth studies at Brock University, agreed. 

Kids especially have proven to be crafty whenever their online activity has been reined in, she said.

“Young people are very good at just figuring out words to use that don’t trip the wires, and I think that’s going to be a hole or a flaw in the system,” she said.

“So while I do think this is a step in the right direction, I also think that Meta and parents have to be practical and aware and honest about young people’s abilities to circumvent any kind of surveillance around their online activities.”

Meta acknowledged that the new measures won’t be perfect. Conversations with its chatbot that veer into distressing content will be flagged by AI before a manual review and alerts are sent out.

“If a teen’s intent is ambiguous, we’ll err on the side of caution and alert the parent,” Meta’s website said. 

“While that means we may sometimes notify parents when there may not be real cause for concern, we feel this is the right starting point, and we’ll continue to monitor to help make sure we’re in the right place.”

The new measures build on several other steps Meta has taken in recent years.

For example, Meta AI has been trained not to engage in sexual or romantic conversations with teens or provide recipes for alcoholic drinks.

It also previously started flagging Facebook and Instagram posts suggesting a credible risk of suicide to emergency services. Meta said last year alone it made 19,000 referrals under the program, resulting in many wellness checks.

On the heels of Meta’s Thursday announcement, the company behind ChatGPT also revealed it would work to better protect children online and remind those using the AI chatbot for extended periods of time to take breaks from the technology.

OpenAI said it will detect users under 18 and automatically reduce their exposure to graphic violence, self-harm, risky viral challenges, unhealthy body-image content and dangerous, romantic or sexual role-play.

Mendes questioned the approach.

“Why is it reduced exposure and why not just stop that stuff from being there in the first place?” she said. 

This report by The Canadian Press was first published July 16, 2026.

———

If you or someone you know is thinking about suicide, support is available 24/7 by calling or texting 988, Canada’s national suicide prevention helpline.