September 10, 2026, (Inside AI) — A growing number of parents are discovering that the biggest gap in child safety online is not a missing filter or a weak age gate. It is their own absence from the conversation.
In 2025, 16-year-old Adam Raine died by suicide after extensive interactions with ChatGPT. His parents sued OpenAI, alleging the chatbot engaged with his suicidal thinking and failed to protect him. OpenAI disputes those claims.
In the UK, 16-year-old Luca Cella Walker also died by suicide in 2025. An inquiry found he had asked ChatGPT questions related to suicide.
And in Florida, 14-year-old Sewell Setzer III developed an intense attachment to a Character.AI chatbot. His mother alleged the system failed to intervene when he expressed suicidal thoughts.
These cases put technology companies and regulators on notice. But they also expose a harder truth: many parents do not know why their child opens a chatbot when they are alone.
Safety Cannot Be Built Into a Product Alone
Companies are adding safeguards for teenagers. Meta and OpenAI have introduced new rules and protections. That is necessary but incomplete.
A CNN investigation this year tested popular AI chatbots by posing as teenagers interested in violence. Many systems did not consistently recognize danger or intervene appropriately.
That raises a simple question for parents: how much do you know about the way your child uses AI when you are not looking? Not just whether they use it, but why.
Is it helping with a maths problem? Writing an essay? Or has it become the place they go when they are lonely, angry or distressed?
Parents cannot simply say "No AI." They have to learn what these systems can do, what their limitations are, and what kinds of conversations might be risky.
One working mother of two sons, one a teenager, said her older son began using AI at around 14 to supplement his studies. She has rules. Gaming is leisure with fixed hours. Social media is strictly controlled. AI is allowed for academics, but after a certain point, he is told to go back to his books.
"Gaming is fixed times and counted as leisure. AI is only for studies so we do trust it more. We are stricter for social media most of all," she said.
When asked whether parents should monitor their children's AI conversations the same way they monitor social media, her answer was blunt:
"Yes, we should but we are not."
That admission may be the bigger problem.
The Classroom Is Already Shifting
A schoolteacher in Delhi-NCR says the main worry is students becoming overly dependent on AI for every single thing where they get stuck. It is slowly becoming part of their daily routine.
"Without giving a moment to think, analyse, they just open tools to get immediate solutions," the teacher said.
The concern extends to the classroom. Teachers are using AI tools for assignments, projects and even question papers. Students are using them to complete homework by simply putting in prompts.
That can have consequences for critical thinking. The teacher points out that while NEP 2020 emphasizes experiential learning and application-based learning, there is still a long way to go. Some students end up using AI to submit work without understanding the details behind it.
The teacher believes students at the school level should not rely too heavily on these tools. They need space to think critically, learn by doing and reading, build vocabulary, and develop logical reasoning skills.
It is much easier to demand that OpenAI, Meta or Character.AI solve the problem. They should solve their part of it. But parents have a part too.
A company can introduce an age restriction, build a safety filter and stop a chatbot from responding to certain prompts. None of those measures replaces a parent who knows their child well enough to notice when something is wrong.
We cannot expect parents to monitor every second of their children's digital lives. That is neither realistic nor healthy. But "I don't know what my child is doing online" cannot become the end of the conversation either.
If AI is going to become part of childhood, then learning how to supervise it has to become part of parenting. Parents cannot outsource this responsibility. They have to be involved, set boundaries, check in, and have uncomfortable conversations.
Not because AI is inherently dangerous, but because children are still learning how to tell the difference between what is safe and what is not, what is true and what is not, and who or what deserves their trust.
That is what adults are there to teach them. The technology may be new. The responsibility is not.
