The use of AI as a personal companion or confidant for kids is creating debate. Photo / 123RF
A confidential helper, or an unreliable manipulator?
Either way, AI has become a source of emotional support and companionship for 23% of New Zealand school children, according to cyber security firm Norton.
“AI’s just exploding everywhere. Kids could be asking questions to ChatGPT or Copilot. There seems to be growth around AI chatbots,” Norton Asia-Pacific managing director Mark Gorrie told the Herald.
“We want kids to be having healthy relationships ... We don’t want kids replacing family and friends with what is an AI tool.”
Gorrie said most parents probably wanted children to develop critical thinking abilities and avoid harmful content and misinformation, and reliance on AI companionship likely undermined that.
He said in some contexts, AI could be useful, but parents needed to have discussions, even if such talk might create tension.
“Parents have concerns about holding their kids back. Kids want to be accepted. Obviously protecting your kids, making sure they’re having healthy online interactions, is still vital.”
Gorrie said 30% of Kiwi parents already checked their child’s devices, such as by reviewing app usage, settings and installed apps.
The Norton Connected Kids survey found the average baby boomer got their first mobile phone at age 41 but Gen Z kids born from the late 1990s through to the early 2010s did so at age 14.
Norton said 34% of parents surveyed in late April and early May felt AI was not beneficial for children’s learning or creativity.
However, only 41% of Kiwi parents said they had discussed AI dangers such as deepfakes and misinformation with their children.
Elon Musk’s AI chatbot Grok last month was found to have struggled with verifying already-confirmed facts, analysing fake visuals and avoiding unsubstantiated claims.
The Digital Forensic Research Lab (DFRLab) of the Atlantic Council analysed about 130,000 posts in various languages on the platform X before reaching those findings.
In January, the US Federal Trade Commission approached the country’s Department of Justice over a complaint that Snapchat’s AI chatbot harmed young users.
In May, an OpenAI technical report cited in New Scientist said some new AI large language models had higher hallucination rates than the company’s previous “o1″ model introduced last year.
AI hallucinates when it makes up answers to questions, producing false or absurd responses.
Vaughan Couillault, Papatoetoe High School principal, said AI had good and bad uses, just as many technologies had.
He said the issue of young people using cellphones was nuanced – and his school had a useful app where students could access their timetables and grades.
“We’re increasingly turning to AI to create solutions for us.”
On April 29, the Government’s ban on cellphones in school classrooms took effect, aimed at removing unnecessary disturbances and distractions.
Some groups have lobbied for stricter rules but Couillault said his school used a high-trust model to uphold the ban, which seemed to work.
“I’ve got 1800 kids and I would have maybe 10 to 15 confiscations a day.”
Couillault said parents frequently had no idea what their kids were doing with phones, and attempts to regulate or monitor phone use at home could cause conflict.
“Perseverance, and human connection, is the solution for me.”
He said a bigger issue was who actually owned the data young people uploaded to apps or AI programmes.
He queried the Norton survey’s sample size of 1001 adults, saying he had more kids at his school.
Gorrie said the survey size was realistic for New Zealand, indicative of trends and Norton carried out multiple surveys worldwide.
Of respondents, 13% of parents said their children had been victims of cyber bullying.
But since some parents admitted not knowing much about children’s online lives, and bullying and scams were known to often be under-reported, Gorrie said the true number was probably higher.
Lobby group B416 is among those pushing for social media use to be limited to people aged 16 and over.
Entrepreneur Cecilia Robinson, B416 co-chairwoman, said the new Norton findings confirmed what parents were already seeing.
“When kids as young as 12 are turning to AI for emotional support, it’s a clear sign that we’ve handed over digital spaces to children without the right protections.”
She said New Zealand had no independent regulator for online safety and no legal minimum age for social media access.
Robinson said the current system left too many kids exposed, unsupported and unprotected.
Norton’s survey found 41% of parents surveyed said cyber bullying perpetrators were their child’s classmate or peer.
The company said “trolling and harassment spans numerous platforms” today whereas in the past, children could generally avoid bullies apart from at school.
“Visual-first social media platforms lead the charge,” Gorrie said.
Some children were bullied on multiple platforms.
Of parents who said their kids were bullied, 33% said children were bullied on Snapchat, 33% also on Instagram, 30% on Facebook and 28% on TikTok.
About one-quarter of those parents said their child had been bullied via text messages.
The Norton survey added: “Strikingly, 46% of Kiwi parents say they knew their child was being cyber bullied before their child confided in them.”
Norton said that showed many parents were picking up on cyber bullying warning signs – but 28% had still not spoken with children about staying safe online, leaving them under-prepared when risks escalated.
The survey was conducted for the “Connected Kids” 2025 Norton Cyber Safety Insights Report, with 1001 adults surveyed.