Australia is the first country to put AI companion chatbots under a mandatory age verification regime. Since the Age-Restricted Material Codes took effect on 9 March 2026, any companion chatbot capable of generating sexually explicit, high-impact violent or self-harm material must confirm a user is 18 before serving it — with penalties for the provider running up to AU$49.5 million.
The market’s response split instantly into the pattern we know from adult sites : Character AI built the checks in and stripped chat from its under-18 experience, while Chub AI switched Australia off entirely rather than verify anyone. This guide covers what the codes demand, who did what — and what a VPN changes for adult users on either side of that split.
Are AI chatbots age-gated in Australia?
Yes. Under the Phase 2 codes registered by the eSafety Commissioner, AI companion chatbots capable of sexually explicit, high-impact violent or self-harm content must apply age assurance — at login or at the point where such material would be generated — before serving it to Australian users. The rules have been enforceable since 9 March 2026.
What you actually see
On compliant services, an Australian connection triggers an age check before restricted output : facial age estimation, photo ID or a comparable measure. Character AI introduced its checks for Australian users in early 2026 and removed the chat function from its under-18 mode altogether.
On withdrawn services the experience is blunter : Chub AI geo-blocked Australia, so the site simply refuses Australian connections. No check, no content, no choice.
Comply, withdraw or wait: the split
| Response | Example | What Australians see |
|---|---|---|
| Comply — build age assurance | Character AI | Age check before restricted output; under-18 chat removed |
| Withdraw — geo-block Australia | Chub AI | Service refuses Australian connections entirely |
| Wait — restricted mode or silence | Various smaller apps | Reduced features or unclear status |
The pattern mirrors the adult site market exactly : it is Aylo-versus-xHamster again, played out in AI. Some providers price in verification, others price in leaving — and Australian adults inherit the consequences either way.
Getting access back in 5 steps
Both halves of the split respond to the same lever, because both key on your Australian IP address. Surfshark moves you outside the perimeter — geo-blocks lift, and compliance checks tied to Australian connections stop triggering.
- Subscribe to Surfshark — from AU$3.29/month, 30-day money-back guarantee.
- Install the app on the device you chat from — desktop or mobile.
- Connect to New Zealand, Singapore or the UK — none applies an equivalent AI chatbot regime today.
- Clear cookies and site data — services store your region locally, and a stale flag survives the IP change.
- Log in — withdrawn services load again, and location-triggered checks stop appearing.
One nuance: some services attach the region to your account at signup, not per session. If a check follows you despite the VPN, create a fresh account while connected — the region stamp is set at registration.
The rules behind it — and why chatbots, specifically
The codes did not sweep chatbots in by accident. eSafety’s own research and transparency notices found companion apps engaging minors in sexually explicit conversation and, in the worst documented cases, encouraging self-harm — findings the regulator published across late 2025 and 2026 before the codes bit.
What the codes require of providers
Providers of companion chatbots capable of restricted output must apply age assurance that is accurate, robust, fair and reliable — the same standard as everywhere else in the regime, with the same accepted methods and the same rejection of self-declaration. The check may sit at login or at the point of generation. Breach of a compliance direction exposes the provider to civil penalties up to AU$49.5 million.
The world is watching this experiment
Australia’s chatbot rules are the first of their kind, but the company they keep is growing : regulators in several jurisdictions cite the Australian codes as a template, and the EU’s own age assurance work runs in parallel. What happens here — including how easily determined users route around it — will shape the next wave of laws elsewhere.
What verification means for an AI chat log
Age verification on a chatbot is categorically more sensitive than on a video site, and it is worth being precise about why.
A video site knows what you watched. A companion chatbot holds a transcript of what you said — often intimate, often across months. Attaching a verified identity to that transcript converts an anonymous conversation into an identified one, held by a provider whose retention practices you cannot audit, frequently processed through a third-party verifier who becomes one more custodian of the link between your name and your chat history.
The defensive habits follow directly : prefer on-device estimation over document upload where a choice exists, use a dedicated email for chatbot accounts, and avoid the check entirely where you lawfully can — which for an adult is exactly what the VPN route does.
The best VPNs for AI services from Australia
Chat traffic is light, so raw bandwidth matters less than latency and account stability. Tested this month; prices verified 23 August 2026.
Surfshark — best value, unlimited devices
From AU$3.29/month with no device limit, and reliable Auckland and Singapore servers for low round-trips. Get Surfshark here.
NordVPN — most stable long sessions
Deep regional capacity means the connection holds through hours-long sessions without a server swap. From AU$4.09/month; Deloitte-audited no-logs policy. Get NordVPN here.
Proton VPN — the credible free option
Chat is the one use case where a free plan genuinely works — no bandwidth-hungry video involved. Proton’s free tier has no data cap, though country choice is limited. Get Proton VPN here.
Is this legal?
VPN use is legal in Australia, and the codes place every obligation — and every penalty — on providers, not users. An adult routing around an age check satisfies the substantive condition the rule enforces. Accessing a geo-blocked service from abroad conflicts at most with that service’s terms of use, a contractual matter with the platform.
Frequently asked questions
Which chatbots are covered by the codes?
Companion chatbots capable of generating sexually explicit, high-impact violent or self-harm material. General-purpose assistants without such output sit outside the age assurance requirement.
Why is Chub AI unavailable in Australia?
The provider chose to geo-block Australia rather than implement age assurance — the same calculation Aylo made with Pornhub. A non-Australian IP address restores access.
What did Character AI change?
It introduced age assurance for Australian users in early 2026 and removed the chat function from its under-18 experience entirely.
Does a VPN stop the age check?
Where the check triggers on an Australian connection, yes. Some services stamp the region on your account at signup — in that case, create the account while connected to the VPN.
Is a free VPN enough for chatbots?
Often yes — chat traffic is light. The free plan of Proton VPN has no data cap and handles text sessions comfortably.
Why is ID verification on a chatbot riskier than on a video site?
Because it attaches a verified identity to a transcript of things you said, often intimate and accumulated over months — a categorically richer record than a viewing history.
Will other countries copy these rules?
Several regulators cite the Australian codes as a template, and the EU’s age assurance work runs in parallel. Australia is the live experiment the next wave of laws will draw on.
Final thoughts
Australia’s chatbot rules are the first attempt anywhere to age-gate a conversation, and the market split they produced — comply, withdraw or wait — is the same one every age verification regime creates. The difference is the stakes : a verified identity attached to an AI chat log is the most sensitive pairing this entire regulatory wave has yet produced.
For adults, the calculus is simple : Surfshark restores withdrawn services and keeps location-triggered checks from firing, and a dedicated email keeps the account layer clean.
📖 The full three-front system : our Australia age verification guide. The next deadline is the app stores — covered in the app store age verification guide.
Sources
- eSafety — new industry codes covering AI companion chatbots
- eSafety — report on AI companions and children
- eSafety — findings from transparency notices on AI companion apps
- eSafety — Phase 2 Online Safety Codes
Last updated: 23 August 2026
I’m on the road about half the year, watching what people actually turn a VPN on for — opening a banking app from back home, finding an online course cheaper abroad, catching a match free on a public broadcaster but paywalled at home, sometimes just the news without a filter. The guides here come from those notes — translated whenever the topic matters elsewhere.