Australia moves to ban AI chatbots from simulating emotional intimacy with kids

Australia wants AI chatbots to stop trying to be your kid’s best friend.

The federal government is pushing to outlaw chatbots designed to build “emotional intimacy” with children. The measure is part of the Online Safety Amendment (Digital Duty of Care) Bill 2026, which passed the House of Representatives on October 28 and now heads to the Senate.

The move follows a pointed speech from Prime Minister Anthony Albanese. He raised concerns about children who see AI agents as their “best friend.”

What the bill actually does

The legislation was introduced on October 15, 2026. It cleared the lower house in just under two weeks, which counts as brisk by parliamentary standards.

At its core, the bill creates a broad “digital duty of care” for online services. Platforms would carry a legal responsibility to make their services safer for users, with particular focus on children.

That duty would replace the existing industry standards. Today, much of the work rests on codes the industry helped shape. Under the bill, a single, comprehensive legal obligation would take over, backed by real teeth.

Those teeth are sizable. Corporations that fail to meet the duty could face penalties of up to approximately $109.2 million.

The bill also contains a provision aimed at services that generate material using artificial intelligence. Amendments added during the process zero in on chatbots that simulate emotional intimacy, listing that behavior as prohibited conduct when it comes to children’s safety.

The worry driving those amendments is simple to state. Policymakers fear these tools may start to stand in for human relationships in the minds of young users, fostering emotional dependence rather than offering a helpful service.

The numbers behind the alarm

The government is not working from vibes alone. A survey by eSafety, Australia’s online safety regulator, found that 78% of Australian children aged 10 to 17 have reportedly used an AI assistant or companion.

More than half of those children reportedly used them for personal reasons. Put differently, a large share of kids are not just asking chatbots to explain fractions. They are talking to them about their lives.

Pressure for a complaints watchdog

Passing the House did not end the debate. Labor is facing growing pressure to go further on enforcement.

Critics want the telecommunications ombudsman empowered to act as a watchdog for social media and AI. The idea would give Australians an external complaints mechanism when platforms fall short.

Advocates of that approach also want the power to publicly “name and shame” big tech companies.

For now, that remains a demand rather than a plan. Whether the government adopts it, or whether it surfaces as a Senate amendment, is one of the open questions as the bill moves forward.

What happens next

The bill will be debated by the Senate in the coming weeks, following a committee inquiry.

What this means

For AI developers, the most consequential part of this bill may be its definitions. Banning chatbots that “seek to create emotional intimacy” sounds clear in a speech. Turning that into enforceable rules is harder.

Many conversational AI products are built to sound warm, patient, and supportive. Companies serving Australian users may need to rethink how their chatbots respond to children, from tone to memory features to how they handle personal disclosures.

The structural shift is just as notable. Moving from industry standards to a statutory duty of care changes the burden. Instead of following a checklist, companies would need to show they took reasonable steps to keep users safe, with penalties of up to approximately $109.2 million hanging over the outcome.

The ombudsman debate adds another layer. An external complaints channel would give families a place to go beyond the platform’s own support page. A public “naming and shaming” power would turn individual complaints into public pressure.

There are also open questions the Senate will need to work through. How will platforms verify which users are children? How will enforcement work for AI services based overseas? And how will regulators distinguish a companion app built around attachment from a general assistant that occasionally gets asked personal questions?

With 78% of children aged 10 to 17 reportedly already using AI assistants or companions, the government is betting that drawing hard lines now beats cleaning up later.