Following the world-first ban on social media for under-16s that took effect in December 2025, the Australian government is now applying similar pressure to AI companies.
The March 9 Compliance Deadline
The Australian eSafety Commissioner has issued a final warning to AI providers, with a major compliance deadline set for Monday, March 9, 2026.
- Age Verification: AI services must implement robust systems to verify user ages.
- Restricted Content: Platforms like ChatGPT and Claude must actively prevent users under 18 from accessing content related to pornography, extreme violence, self-harm, and eating disorders.
- Gatekeeper Penalties: If an AI service fails to comply, the regulator has warned it may force app stores (Apple/Google) and search engines to block access to those services within Australia.
- Heavy Fines: Non-compliant companies face fines of up to A$49.5 million ($35 million).
Current Compliance Landscape
A recent review found that many AI platforms are currently “unprepared” for these specific Australian laws:
| Compliance Category | Status (as of March 2026) |
| Fully Prepared | 9 out of 50 top platforms (including OpenAI/ChatGPT and Anthropic/Claude have begun rolling out age controls). |
| Blanket Filters | 11 platforms are applying “blind” filters to all Australian traffic to avoid fines. |
| Non-Compliant | 30 platforms showed no public plans to meet the March 9 deadline and risk being blocked. |
| Voluntary Exit | Some smaller “companion” chatbots (e.g., HammerAI) have chosen to block Australian IP addresses entirely rather than build the required safety tech. |
National AI Plan (2026-2030)
While the eSafety Commissioner is focusing on “harms,” the broader Federal Government strategy is shifting:
- No “New” Dedicated AI Law: In December 2025, the government decided to rely on existing legal frameworks (privacy, consumer law, and online safety) rather than passing a standalone “AI Act” like the European Union.
- AI Safety Institute: Australia is establishing its own AI Safety Institute in the first half of 2026 to monitor “frontier” risks and catastrophic threats.
- Schools Overturn Bans: Notably, while the government is tightening rules for kids’ private use, public schools in NSW and South Australia have recently overturned previous bans on ChatGPT, instead rolling out safe, department-owned versions like “NSWEduChat” for students.
The “Sovereignty” Argument
Assistant Technology Minister Andrew Charlton recently stated that Australia’s goal is to transition from “renting” AI to “owning” it. The government is pushing for more Australian-owned AI infrastructure to ensure the profits and intelligence don’t simply flow offshore to Silicon Valley.
