New Zealand is sticking with its call for a ban on war robots that do not comply with international humanitarian law.
The government met a deadline at the weekend to make a submission to the UN on autonomous weapon systems.
"New Zealand continues to seek binding international prohibitions and regulation of AWS ... [but] New Zealand does not consider that a blanket ban on AWS will be constructive or gain broad support at this time," the submission said.
The world's militaries are increasingly using drones in the air, on land and at sea.
Almost a dozen countries have announced drone swarm programmes, where dozens or more drones coordinate.
Artificial intelligence is increasingly added to the mix. While not all AWS use AI, "rapid recent advances in technologies such as machine learning and AI ... mean that AWS are now potentially much more mobile, versatile, capable and lethal," a briefing to government ministers in January said.
Not all of of the systems are lethal.
"AWS that meet ethical and legal requirements may offer legitimate military benefits," the New Zealand submission said.
The government's position was reconfirmed in January, along the lines of what was established three years ago.
"We hope the international community can take the right turn at this crossroads," it told an AWS conference in Vienna last month.
It has also signed up to a new arms control push around AI, the first "Political Declaration on Responsible Military Use of Artificial Intelligence and Autonomy" led by the US.
"Responsible militaries will develop and use AI-enabled capabilities in innovative ways to strengthen protections for civilians in armed conflict, improve decision-making and logistics, and enhance humanitarian and disaster relief," the US said in issuing it.
The Defence Force uses AI for automated deployment of decoy countermeasures under attack by torpedoes and missiles, and for simulations for research and training.
The navy is trialling a drone called 'Blue Bottle'.
The January briefing to New Zealand ministers said the country needed to preserve access to technology "relevant for our national secuirty and interoperability with security partners".
"AI gives users significant advantages, including greater speed, efficiency, accuracy and situational awareness."
But this had to be accompanied by human oversight to ensure laws were applied around differentiating between civilians and military targets and proportionality, and holding humans to account for violations, it said.