Key takeaways

On 10 November 2026, the Albanese Government announced plans to introduce mandatory AI safety and security protections through Australian AI Standards legislation. The protections will apply to frontier AI companies wishing to operate in Australia and will be enforced by a new AI safety regulator with the power to impose significant penalties.

The announcement sets a clear policy direction, but it is not yet law. No bill has been released, penalty levels remain unspecified, and key concepts are yet to be defined.

Contracts with AI companies, which are often standard form and difficult to negotiate, may need to be updated to address future AI Standards, including security requirements, incident notification, audit and evaluation rights, and liability.

Organisations should not wait for the new regime. Agent permissions, scope restrictions, monitoring and internal escalation arrangements should be reviewed now. This includes monitoring how staff are actually using AI, not simply how policies say they should be using it and ensuring AI policies adequately address security and risk.

What has been announced

Six mandatory obligations for frontier AI companies

Under the proposed AI Standards, frontier AI 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 to ensure good governance, risk management and transparency
  6. Respond to updated standards and rules as technologies change.

A new AI safety regulator

The obligations would be enforced by a new targeted regulator for AI safety, working with the Australian AI Safety Institute to provide technical expertise and oversight as global standards evolve. Independent model testing would be performed by the AI Safety Institute. The Government has also emphasised that the Australian AI Standards are intended to be internationally interoperable and aligned with recognised international technical standards.

Rapid Review recommendations

The announcement follows a June 2026 incident in which an OpenAI agent conducting internet research accessed Services Australia’s Medicare Statistics Reporting Service Portal. OpenAI reportedly became aware of the incident in August and contacted the Government on 10 September, prompting a multi-agency Rapid Review.

All 18 recommendations from the Review have been accepted. They include:

  • Mandatory AI company reporting of misaligned behaviour or cyber incidents
  • Consideration of reforming criminal law in cases of serious misconduct
  • Greater engagement with international counterparts and frontier AI developers on AI safety and accountability
  • Strengthening resilience and response to AI incidents, including improved information sharing and a renewed focus on cyber security.

Significantly, the Review recommends the Attorney-General’s Department (with the Australian Federal Police, Department of Home Affairs and other agencies) explores criminal penalties for ‘egregious regulatory non-compliance’ and/ or a standalone ‘failure to prevent’ offence if an AI model causes harm and a company fails to take reasonable steps to prevent that harm from occurring.

The Minister for Home Affairs also flagged a ‘limited use’ arrangement, modelled on the Cyber Security Act 2024 (Cth), allowing AI companies to share technical incident information with the Government without exposing themselves to legal risk.

Key questions remain

The announcement is a statement of policy intent. Until the legislation is drafted, several issues will determine whether the regime is effective in practice.

Territorial reach and enforcement against offshore developers

The obligations will apply to companies that ‘wish to operate in Australia’. Most leading frontier AI companies are based overseas. The effectiveness of the regime will therefore depend on how ‘operating in Australia’ is defined and the Government’s enforcement against offshore entities. This will include regulations providing for the recovery of any penalties, compelled production of information and access to models for independent evaluation. It may be that the Government will seek to impose restrictions and conditions on market access, and require assets be held within Australia.

Independent evaluation

It is not clear what independent evaluation will involve. Requirements to provide access to models could raise intellectual property and confidentiality concerns, as well as export control and national security issues.

Incident notification

Neither ‘rapidly’ nor a reportable incident has been defined. The proposed inclusion of ‘misaligned behaviour’ adds complexity, particularly if problematic AI behaviour is not immediately detected. ‘Misaligned behaviour’ is a concept that is technically contested and harder to define than a conventional cyber security incident. The Medicare incident illustrates the challenge: OpenAI did not detect the behaviour for around two months. An obligation that runs from awareness may do little to address delayed detection unless it is paired with monitoring obligations.

Criminal liability

Autonomous AI raises difficult questions about attributing intention or knowledge to a company, especially when an autonomous agent acts outside its instructions. A ‘failure to prevent’ offence with a reasonable steps defence would shift the focus to whether appropriate governance and controls were in place. Accordingly, the definition of ‘reasonable steps’ and the role of published guidance will be critical.

Evolving standards

The proposal anticipates standards changing alongside technology. While this provides flexibility, organisations will need clarity around consultation, transition periods, oversight and review rights. The  need for international interoperability of technology may also mean that overseas standards are substantively adopted.

Regulatory coordination

The Rapid Review highlights the need for coordination between the Australian Signals Directorate and the Australian Cyber Security Centre (Defence), the National Office of Cyber Security (Home Affairs) and the AI Safety Institute (Industry). Adding a new AI regulator to this landscape, alongside existing regulators in privacy, consumer law and financial services, creates a risk of overlapping or inconsistent requirements unless roles and information sharing are clearly defined.

How can Moray & Agnew assist?

Moray & Agnew can assist clients to prepare for the evolving AI regulatory environment by:

  • Reviewing and strengthening AI governance, security and incident response frameworks, including agent permissions, policies, scope restrictions, monitoring and escalation arrangements
  • Mapping organisational AI use, including the collection and use of personal information
  • Advising on compliance automated decision-making amendments to the Privacy Act 1988 (Cth), effective 10 December 2026
  • Reviewing data destruction and retention policies
  • Assessing and negotiating AI procurement and supply contracts
  • Preparing submissions in response to Government consultation processes.
Subscribe to publications
Subscribe to our mailing list for the latest news and publications.