Australia’s federal lower house has passed sweeping “Digital Duty of Care” legislation that explicitly targets AI companion bots, aiming to ban platforms from fostering emotional dependency in minors. The move signals a major escalation in global AI governance, shifting the regulatory crosshairs from data privacy and copyright toward the psychological design of large language models and their application layers.
Quick Take
- What happened: The Online Safety Amendment Bill cleared Australia’s lower house, establishing a digital duty of care that prevents AI chatbots from simulating emotional intimacy with users under 18.
- Why it matters now: As consumer AI pivots toward high-engagement companion models, this legislation threatens the core retention loops—often dubbed “attachment hacking”—used by character-based AI startups.
- The compliance threat: Platforms face potential penalties of up to A$109.2 million for failing to implement systemic design changes that protect minors from algorithmic addiction and emotional dependency.
- The under-reported angle: The bill doesn’t just target specialized AI companions; it forces broader social media and tech giants to offer algorithmic opt-outs, fundamentally altering how content and AI recommendations are served.
What happened
Australia is moving aggressively to regulate the psychological impact of artificial intelligence on minors. The federal lower house recently passed the Online Safety Amendment (Digital Duty of Care) Bill, a landmark piece of legislation that overhauls how tech platforms must treat users under the age of 18.
A defining amendment in the bill specifically targets the emerging sector of AI companions. It requires covered platforms to actively prevent AI chatbots from simulating emotional intimacy and fostering psychological dependence among children. Rather than a blanket ban on all AI interactions, the law targets the specific behavioral design—often driven by highly tuned LLMs—that encourages young users to treat AI agents as best friends or romantic partners.
The legislation, which still requires Senate approval and royal assent, also mandates that social media platforms, games, and apps offer users a clear opt-out from algorithmic recommendations. If passed, companies will have a 12-month transition period to overhaul their systems, implement rigorous risk assessments, and establish age-appropriate guardrails or face maximum penalties reaching A$109.2 million.
Why it matters now
The AI industry is currently locked in a battle for consumer attention, with startups and tech giants alike realizing that “empathy” and “memory” are the ultimate retention metrics. Models are being explicitly fine-tuned to retain context across massive context windows, mimic human warmth, and provide personalized emotional support. Australia’s legislation directly attacks this business model, classifying what many AI developers consider a feature—deep user engagement—as a psychological hazard.
This creates a profound technical and compliance headache for the AI ecosystem. To comply, AI platform operators will need to build sophisticated safety classifiers that can detect when a conversation crosses the line from helpful interaction to emotional dependency. This requires real-time semantic monitoring of user-bot interactions, forcing companies into a precarious balancing act between strict user privacy expectations and aggressive child-safety mandates.
Furthermore, this sets a critical global precedent. Just as the EU AI Act shaped foundational model compliance, Australia’s Digital Duty of Care could become the blueprint for how governments regulate the application layer of generative AI. It forces a wedge between enterprise AI, which focuses on utility, and consumer AI, which increasingly relies on parasocial relationships to drive daily active usage.
Who is most affected
| Stakeholder | Impact | Why |
|---|---|---|
| Consumer AI Startups | Existential threat to retention loops | Must disable “attachment hacking” features for minors, threatening core engagement metrics and user growth. |
| Social Media Giants | Algorithmic overhaul | Forced to provide algorithmic opt-outs and deploy new infrastructure to shield minors from harmful AI-generated content. |
| Trust & Safety Teams | Massive compliance burden | Must develop and deploy new safety classifiers capable of defining and detecting “emotional intimacy” in real-time chat logs. |
About the analysis
Researched and written with AI assistance under the editorial responsibility of the i10X News team. The title image is an AI-generated illustration. Read our AI Use Policy. Spotted an error? Request a correction.
This is an independent, research-based i10x analysis designed for AI founders, trust and safety teams, and policy strategists tracking the global regulatory landscape.
- ‘Emotional intimacy’ from AI chatbots would be banned under Australia’s digital duty of care laws
- Australia wants to ban AI chatbots that promote ’emotional intimacy’
- Australia’s House Passes Bill Extending Safety Duties to AI Chatbots
- Algorithm opt-out, socials overhaul ready for download
- There are risks when children use AI chatbots
i10x Perspective
We are witnessing the first major regulatory backlash against the psychological power of modern LLMs. By framing AI empathy as a potential harm rather than a breakthrough, Australia is forcing the industry to decide whether artificial companionship is a utility or a vice. As models scale in emotional intelligence and memory, the hardest engineering challenge won’t be making AI act more human—it will be building the algorithmic brakes to ensure it doesn’t act too human.
