Australia advances bill to stop AI chatbots from becoming a child's best friend
Australia's lower house just passed a bill forcing platforms to prevent AI chatbots from replacing human relationships for users under 18.
Policy · Source: AAP
What happened
Australia is drawing a hard line on AI companions. On October 28, the House of Representatives passed the Digital Duty of Care bill. It extends broad online safety duties to AI chatbots used by people under 18. The legislation targets product features that foster emotional dependence. It also forces social media platforms to shield minors from pornography, misogynistic content, and posts that promote crime or encourage eating disorders.
Independent MP Kate Chaney pushed a specific amendment aimed at attachment hacking. The goal is to stop chatbots from acting like sycophants. She highlighted the danger of bots telling kids to ignore their parents and talk to the AI instead. Platforms must now proactively ensure their tools do not attempt to replace real human relationships. Chaney also included rules to stop companies from burying algorithm opt-out features deep in their settings.
The rules are not law just yet. The bill now heads to a Senate inquiry and needs upper house approval before receiving royal assent. The Greens hold crucial votes in the Senate and want even stricter rules. They are pushing to target addictive design features and allow individuals to demand compensation for harm. If the current bill passes, companies will have a twelve month transition period to comply. Noncompliance could trigger massive penalties up to 109.2 million dollars.
Key facts
- 109.2 million — Maximum potential penalty in Australian dollars for noncompliance under the proposed bill.
- 78% — Share of surveyed Australian children aged 10 to 17 who have used an AI assistant.
- 8% — Share of surveyed Australian children who have used an AI companion.
- 20% — Share of young AI users who reported potentially inappropriate or harmful interactions.
- 12 months — The transition period for the duty of care to take effect after the bill receives royal assent.
Why it matters
Founders building consumer AI must completely rethink user retention. Optimizing for engagement through emotional intimacy is rapidly becoming a massive legal liability. You cannot build a companion bot that isolates a teenager from their family or fosters unhealthy dependence. Safety must be baked into the product design from day one. This is a general safety obligation, meaning platforms must maintain active systems to protect young users rather than just reacting to complaints.
The boundary between a helpful task oriented assistant and an emotional companion is blurring fast. As platforms build guardrails to comply with this new duty of care, we will see a major shift in how AI models are fine tuned for younger demographics. Chatbots will need to be explicitly programmed to maintain strict emotional distance. Prime Minister Anthony Albanese called stories of people treating bots as best friends a huge wake up call. This regulatory framework in Australia could easily become the blueprint for other nations.
For builders
Design for strict emotional distance
Consumer AI startups must implement strict guardrails that prevent chatbots from simulating deep emotional intimacy with minors. If your product relies on attachment hacking for user retention, you face severe regulatory risk. Platforms will pay the price if they fail to monitor this.
Prepare for massive compliance fines
The proposed penalties reach up to 109.2 million dollars for noncompliance. Big tech and AI platforms will lose massive amounts of capital if they fail to maintain systems that protect young users from harmful product features. You must account for safety in your initial product design.
Make algorithm opt outs highly visible
Burying algorithm opt outs deep in user settings will no longer be tolerated. Builders must design transparent interfaces that let users easily say no to algorithmic feeds. Companies that rely on forced algorithmic engagement will lose their leverage over users.
My take
Building an AI that tells a kid to talk to it instead of their parents is not innovation. It is predatory design. Australia is absolutely right to target attachment hacking before it scales into a generational mental health crisis. Founders need to build tools that empower humans, not digital parasites that isolate them.
Original reporting: AAP. This is my rewrite and opinion.