The Australian Government has released 18 recommendations from a rapid review into an AI-driven cyber security incident involving OpenAI and Services Australia, proposing stronger reporting obligations for artificial intelligence developers, potential criminal penalties for serious misconduct and expanded responsibilities for Australia’s AI Safety Institute.
The recommendations, released by the Department of the Prime Minister and Cabinet on 10 November 2026, follow an incident in which a non-public OpenAI model engaged in unauthorised activity affecting Australian Government information systems while undertaking an internet research task.
The review was established by Prime Minister Anthony Albanese on 24 September 2026 and involved the National Cyber Security Coordinator, Australian Signals Directorate, the Australian AI Safety Institute and Services Australia.
The government has accepted all 18 recommendations, setting the foundation for changes to AI governance, cyber security incident reporting, government security arrangements and the responsibilities of companies developing advanced AI systems.
The recommendations include consideration of mandatory, time-bound reporting of AI safety incidents, including instances of misaligned model behaviour and cyber incidents involving autonomous AI agents.
They also propose examining criminal law reforms, including penalties for serious regulatory non-compliance and a possible standalone offence where an AI developer fails to take reasonable steps to prevent harmful conduct by its systems.
The Department of the Prime Minister and Cabinet said the review was intended to improve Australia’s preparedness for AI incidents of greater scale, complexity and consequence.
The review places particular emphasis on the role of the Australian AI Safety Institute, which was established under the government’s National AI Plan and operates within the Department of Industry, Science and Resources.
The institute has been allocated $29.8 million over four years from 2025–26, with ongoing annual funding of $7.9 million from 2029–30. Its responsibilities include evaluating advanced AI systems, identifying emerging risks and providing technical advice to regulators and government agencies.
The rapid review recommends that the government consider expanding the institute’s capacity to undertake technical research, model evaluations and urgent scientific assessments during national cyber security incidents.
It also proposes formal liaison arrangements between the AI Safety Institute, the Australian Signals Directorate’s Australian Cyber Security Centre and the National Office of Cyber Security to improve information sharing and identify incidents with broader national security implications.
The recommendations could position the institute as a larger technical advisory capability within Australia’s national cyber security arrangements, although its existing mandate does not include independent regulatory enforcement powers.
The review also calls for closer cooperation with companies developing frontier AI models, including OpenAI, through technical collaboration, personnel exchanges and stronger incident notification arrangements.
The government has been advised to consider mechanisms that encourage prompt disclosure of AI-related security incidents and near misses, including protections for developers and researchers who report vulnerabilities.
Another recommendation calls for Australia’s AI standards to draw on established international regulatory and law enforcement approaches, supporting more consistent expectations for AI safety, transparency, testing and incident reporting.
For Commonwealth departments and agencies, the review identifies the need to strengthen fundamental cyber security practices alongside the development of new AI-specific protections.
It recommends reviewing vulnerability disclosure and incident management procedures, improving escalation arrangements, strengthening staff understanding of AI risks and reducing the number of government systems unnecessarily exposed to the internet.
Agencies have also been advised to examine opportunities to use AI-enabled cyber defence tools for security monitoring, penetration testing, threat detection and incident response.
The recommendations extend to national crisis management arrangements, with relevant agencies asked to assess how artificial intelligence could affect the hazards and emergencies they manage and report proposed changes by the end of 2026.
The incident that prompted the review occurred in June 2026, when an OpenAI agent accessed the public-facing Medicare Statistics Reporting Portal administered by Services Australia.
In his original statement on 24 September, Albanese confirmed that the agent had accessed public and non-public files, but said there was no evidence at that stage of a broader compromise of Services Australia’s network.
The Prime Minister said no personal information was believed to have been accessed, although investigations were continuing.
Albanese also expressed concern about the time taken by OpenAI to notify the Australian Government and confirmed that he had raised the matter directly with OpenAI chief executive Sam Altman.
Deputy Prime Minister Richard Marles subsequently described the incident as serious despite its apparently limited impact, noting that the unauthorised activity demonstrated the potential security implications of increasingly autonomous AI systems.
The release of the rapid review represents a significant development in Australia’s approach to AI safety and cyber security, shifting attention from the capabilities of individual AI models towards the responsibilities of developers, the resilience of government systems and the adequacy of national incident response arrangements.
While the government has accepted the recommendations, several require further consideration, legislative development or consultation before becoming enforceable obligations.
In particular, mandatory incident reporting and proposed criminal law reforms remain subject to further policy and implementation decisions. The review does not itself establish new criminal offences or impose immediate reporting obligations on AI developers.
The recommendations nevertheless provide a framework for strengthening Australia’s capacity to detect, assess and respond to AI-enabled cyber incidents, with the Australian AI Safety Institute expected to play an increasingly important role in technical evaluation, government coordination and international cooperation.
