Artificial intelligence (AI) is playing a central role in the ongoing war in the Middle East. The United States, for example, has acknowledged using the technology to identify potential targets and accelerate decision-making.
This is part of a growing trend. And in some cases, this has led to an increase in the number of civilian deaths.
Against this backdrop, the Australian Department of Defense has announced a new AI policy.
This policy aims to govern the use of AI by the Australian military. So what does it include? And how does it differ from other countries’ military AI policies?
Three main requirements
Australian policy sets out three overarching requirements for the Department of Defense’s use of AI.
First, the use of AI must comply with Australian law and international obligations.
Second, the use of AI must be underpinned by personal responsibility and consider the impact on people. It must also be explainable, reliable, secure, and designed to reduce unintended bias and harm.
Third, risks associated with the use of AI must be managed with appropriate controls such as testing, training, and evaluation.
It is noteworthy that this policy focuses on proportional control.
AI is not a separate item. It is a technology that enables many applications that can be incorporated across a variety of military functions, including targeting, logistics, training, and maintenance, each of which poses different risks.
The policy aims to cover the full range of AI technologies, from chatbots to cutting-edge “frontier” general-purpose AI models.
This approach reflects the Australian Government’s Responsible Use of AI Policy, which came into effect in September 2024.
This policy clearly separates the defense portfolio from the national intelligence community. New policies will fill that gap.
Not paying attention to details
The policy says little about how the Army, Navy, Air Force or other defense agencies, such as the Australian Strategic Capabilities Accelerator, would actually enact its requirements.
It also said that testing and evaluation of the use of AI by the Department of Defense will be an important control tool. However, there are no details on how this will be implemented for military AI. This is an area where testing poses well-documented challenges regarding unpredictable behavior and unreliable performance in military operational environments.
Established in 2024, the Defense AI Center is recognized as a governance hub. However, this policy does not focus much on implementation, compliance, monitoring, resources, or reporting.
It remains to be seen how these settings will evolve and whether guidance on their implementation will follow and be published.
Based on precedent
Australia’s policies are informed by those of its closest allies.
For example, the UK adopted a Defense AI Strategy in 2022 and published the Trustworthy AI in Defence Directive in 2024.
The UK is also working to appoint “Responsible AI” officers within each department of the Ministry of Defence. A progress report for 2025 was also released.
In 2020, the U.S. Department of Defense adopted AI Ethical Principles. After two years, we developed a detailed implementation strategy. And in January 2026, the current administration released the Department of the Army’s AI Strategy. This shifted the focus to speed and lethality, mandated the “legal use” (not necessarily consistent with ethical use) of AI, and directed the removal of barriers to rapid adoption.
Australia’s defense AI policy is generally aligned with these core elements of a like-minded military. This means AI must be used legally, humans must remain responsible, and risks must be anticipated, avoided, and mitigated.
One notable difference in Australia’s policy is the reference to Article 36 of Additional Protocol I to the Geneva Conventions. The policy mandates legal review of AI in weapons systems, which few countries have enacted.
Another difference is that Australia’s policy lacks the implementation roadmap found in US and UK policies. Rather, it reads like a statement of intent.
It is unclear what impact, if any, this change in policy and institutional depth is likely to have on AUKUS Pillar II, which includes cooperation on the acceleration and rapid integration of AI and autonomous technologies.
The growing importance of national frameworks
International efforts to control military AI may be losing momentum. Multilateral discussions on autonomous weapons have also stalled.
This means that national policy frameworks will take on greater importance, shaping procurement and informing partners of what countries consider acceptable practices.
The contemporary use of military AI in ongoing conflicts such as Iran, Lebanon, Gaza, and Ukraine is a reminder that governance is not an abstract policy practice.
Australia’s new policy settings are an important step. The test will be whether implementation measures are strong enough to effectively manage the development and use of military AI.
