TL;DR
Australia's Ghost Bat autonomous drone fired an air-to-air missile in a live test over South Australia, marking a milestone in AI-assisted lethal autonomy while revealing the deliberate limits the ADF is placing on machine decision-making.
What happened
- Ghost Bat drone completed a live air-to-air missile shoot-down over remote South Australia in the weeks before Christmas 2025, with no human in the cockpit or at a joystick.
- The aircraft runs on deterministic programming, not open-ended AI: thousands of pre-designed options are loaded, and the system selects among them with predictable, auditable outcomes.
- An operator aboard an E7 Wedgetail aircraft issued the engagement order and gave final launch authorization before the missile fired.
- Defence Minister Richard Marles separately traveled to the US to meet OpenAI, Anthropic, Microsoft, and Google on where their models are heading for defence use.
- The ADF is also deploying Palantir's Maven Smart system for intelligence analysis, a move that has drawn scrutiny over founder political ties and data sovereignty concerns.
Why it matters
- Human-machine teaming is now operational, not theoretical: Australia joins a short list of nations that have live-tested AI-adjacent lethal autonomy in a controlled but real-world scenario.
- The deterministic architecture is a deliberate legal and ethical hedge: Air Marshal Stephen Chappell says full confidence in outcomes is only possible because the system cannot improvise, a direct response to international pressure on autonomous weapons.
- The UN and ICRC issued a joint call to ban autonomous weapons that are unpredictable or that target humans, framing Australia's cautious design choices as a geopolitical signal as much as a technical one.
- Israel's use of AI tools including the reported Lavender system, which generated a list of 37,000 potential targets, shows the high-stakes downside: the IISS says faster targeting has increased error risk, not reduced it.
- The bigger near-term impact may be upstream of weapons entirely: analysts like ANU's Aina Turillazzi argue AI's real military leverage is in assembling the intelligence picture before any trigger decision is made.
What to watch next
- Whether Australia formalizes a human-authorization doctrine for autonomous weapons, which would set a precedent other Five Eyes partners could adopt or pressure-test.
- Expansion of the Ghost Shark autonomous submarine program, the underwater counterpart to Ghost Bat, as the next visible milestone in ADF autonomous systems.
- How the Palantir Maven Smart deployment resolves data sovereignty concerns, since the outcome will shape which commercial AI vendors gain long-term access to allied defence infrastructure.
Originally published on Present of AI, a daily source-linked AI news timeline. Read the full timeline or browse the open dataset.