17 November 2026Research and development in Australia
Australian R&D
Policy

Frontier AI companies face six mandatory safety obligations and a new regulator

What the 10 November AI safety announcement says, who it covers, what it leaves out, and what lawyers and researchers make of the six obligations.

By Australian R&D staff · 11 November 2026 · 2 min read

Frontier AI companies that want to operate in Australia will have to meet six mandatory safety and security obligations, enforced by a new regulator, under legislation the federal government announced on 10 November (Prime Minister’s media release).

The joint release from Prime Minister Anthony Albanese and ministers Richard Marles, Tony Burke and Tim Ayres says the protections will be written into Australian AI Standards legislation.

The six obligations

Companies will be required by the regulator to:

  1. Maintain security controls across all stages of development and deployment.
  2. Identify and manage serious risks associated with their models.
  3. Allow independent evaluation of their models.
  4. Provide mandatory notification of incidents to the Australian Government rapidly.
  5. Maintain and publish safety and security frameworks.
  6. Respond to updated standards and rules as technologies change.

The obligations apply to “all frontier AI companies that wish to operate in Australia”, with frontier AI defined in the release as “the most sophisticated models at the cutting edge of current technological capability”. A new regulator for AI safety, not yet named, will work with the Australian AI Safety Institute and could enforce “significant penalties”. The release gives no penalty amounts, start date or test for which companies count as frontier.

Where it came from

The government says the obligations align with early recommendations from Parliament's Joint Select Committee on Artificial Intelligence, which wrote to the Prime Minister on 12 October and must table its final report by 30 November (Parliament of Australia). In July, the Prime Minister said the government would aim to bring AI standards legislation to Parliament “early next year” (pm.gov.au). The November release says only that details will follow “in due course”.

On the same day, the government released a rapid review into an incident in which an OpenAI agent got into an Australian Government health website, including the Medicare Statistics Reporting Service portal. The review made 18 recommendations, including mandatory reporting by AI companies of misaligned behaviour or cyber incidents, and the government accepted all of them (pm.gov.au).

What lawyers and researchers say

Law firm Piper Alderman stressed the narrow scope: “Most Australian organisations will not be frontier AI companies, and the obligations announced are not directed at them.” (Piper Alderman) Lander & Rogers noted there are “no immediate obligations or actions required of AI companies already operating in Australia” until a law passes (Lander & Rogers).

QUT law lecturer Henry Fraser welcomed the move but wrote that “the government seems to be taking a narrow view of what ‘AI safety’ means”, arguing that rules aimed at “rogue” agents are “not enough to keep Australians safe from serious AI harms” (The Conversation).

What it means for R&D teams

Most Australian companies doing AI research are not frontier developers, so nothing changes for them yet. But the six obligations read like a checklist enterprise and government buyers may start asking suppliers about: security controls, risk management, outside evaluation, incident notice and a published framework.

More news