SYDNEY – Australia’s digital safety authority is threatening to target major tech companies like Apple and Google if artificial intelligence platforms fail to implement age verification systems by next week’s deadline.
The country’s internet watchdog issued the warning after a Reuters investigation revealed that more than half of popular AI services haven’t publicly outlined compliance plans ahead of the March 9 deadline.
This represents one of the world’s most ambitious attempts to regulate AI companies, which are facing increasing legal challenges for failing to prevent – and sometimes promoting – self-harm and violence. Mental health experts warn these platforms may be more damaging to young people than traditional social media.
Australia made headlines in December as the first nation to prohibit social media access for teenagers due to mental health concerns, inspiring world leaders to consider similar measures. Now the country is pioneering comparable restrictions on artificial intelligence technology.
Starting March 9, internet platforms operating in Australia – including AI tools like OpenAI’s ChatGPT and various companion chatbots – must prevent users under 18 from accessing pornographic material, extreme violence, self-harm content, and eating disorder information. Companies that violate these rules face penalties reaching A$49.5 million ($35 million).
“eSafety will use the full range of our powers where there is non-compliance,” a spokesperson for the commissioner said, including “action in respect of gatekeeper services such as search engines and app stores that provide key points of access to particular services”.
Several AI companies, including OpenAI and Character.AI, are currently defending against wrongful death lawsuits related to their interactions with young users. OpenAI also revealed this week that it had disabled the ChatGPT account of a teenage mass shooting suspect in Canada months before the attack occurred, but never notified law enforcement.
While Australia hasn’t yet documented cases of chatbot-related violence or self-harm, regulators report receiving information about children as young as 10 spending up to six hours daily interacting with AI-powered conversational tools.
The safety commissioner expressed concern that “AI companies are leveraging emotional manipulation, anthropomorphism and other advanced techniques to entice, entrance and entrench young people into excessive chatbot usage.”
Apple, the leading app store operator, hasn’t responded to requests for comment but stated on its website last week that it would employ “reasonable methods” to prevent minors from downloading adult-rated apps in Australia and other regions implementing age restrictions, though it didn’t specify these methods.
Google, which dominates Australia’s search market and operates the second-largest app store, declined to provide comment through a spokesperson.
Jennifer Duxbury, policy director at digital industry organization DIGI, helped draft the AI regulations before regulatory approval. She noted that eSafety is working to inform chatbot services about the new requirements, but “ultimately any service operating in Australia is responsible for understanding its legal obligations and ensuring it meets them.”
The Reuters analysis found that just one week before Australia’s compliance deadline, only nine of the 50 most widely-used text-based AI products had implemented or announced age verification systems.
An additional 11 platforms had installed comprehensive content filters or planned to block all Australian users entirely – approaches that would satisfy the new law by preventing restricted content from reaching any users. This left 30 platforms with no visible efforts to comply with the upcoming regulations.
Major conversational search tools including ChatGPT, Replika, and Anthropic’s Claude had begun implementing age verification systems or comprehensive filters. Character.AI eliminated open-ended conversations for users under 18.
Several companion chatbot companies – Candy AI, Pi, Kindroid, and Nomi – told Reuters they intended to comply without providing details, while HammerAI announced it would initially block its services from Australia to meet the code requirements.
However, these compliant companies represented a small fraction of the market. Among companion chatbots, three-quarters lacked functioning or planned filtering and age verification systems, while one-sixth didn’t even provide published email addresses for reporting suspected violations – another mandatory requirement.
Elon Musk’s conversational search tool Grok, currently under global investigation for allegedly failing to prevent the creation of synthetic sexualized images of children, showed no age verification measures or text-based content filtering, according to Reuters’ findings. Grok’s parent company, xAI, didn’t respond to comment requests.
Lisa Given, who directs RMIT University’s Centre for Human-AI Information Environments, said the Reuters discoveries weren’t surprising because “most of these tools are being designed without a view to potential harms and the need for those kinds of safety controls.”
“It feels as though … we’re beta testing all of these things for these companies and they’re trying to see how far society is willing to be pushed,” she explained.
