Australia

• 2026-10-09: the PM&C consultation on AI-infrastructure standards closed.
• 2026-10-08: Minister Andrew Charlton said the government is finalising systems-based frontier-AI national standards, due by the end of 2026 (proposed).
• 2026-10-06: Joint Select Committee on AI hearings began; OpenAI was questioned over a September incident in which AI agents accessed Australian government sites. This is non-binding scrutiny.

What to watch: publication of the frontier-AI standards by end of 2026, designation of a regulator, and enabling legislation in early 2027.

Australia has no AI statute. The earlier proposal for standalone mandatory guardrails was superseded and never enacted. Current and planned instruments are:

• National AI Plan (2025-12-02). Policy; commits A$29.9 million to the AI Safety Institute.
• Online Safety Act 2021, Phase 2 codes (effective 2026-03-09). Six codes registered 2025-09-09; generative AI is covered mainly under the Designated Internet Services code.
• Social-media minimum-age regime (in force 2025-12-10). AI-adjacent; reasonable-steps age assurance for under-16s.
• NSW AI Operational Policy (July 2026). Mandatory for NSW government agencies.
• Planned: mandatory "Australian Standards for AI" and an Office of AI (announced 2026-07-15; proposed). The announcement has been corroborated by three secondary sources but no primary pm.gov.au release was located, so it is unverified at primary level.

Today, AI is regulated indirectly through technology-neutral law, including online-safety, privacy and consumer law, plus sector rules. The planned frontier-AI standards would be systems-based: companies would have to demonstrate that their safety systems work. A separate set of AI-infrastructure standards, covering data centres and frontier-AI training, has also been consulted on.

Existing exposure arises mainly under the eSafety codes and sector rules; online-safety penalties reach about A$49.5 million. The regulator and the penalties for the planned Australian Standards for AI have not been specified.

Caveats. The 2026-07-15 Office of AI announcement is unverified against a primary source. The date of an eSafety AI-companion notice is unverified, and a report that the 2024 Voluntary AI Safety Standard was superseded on 2026-09-10 is also unverified. Research assistance, not legal advice.