Autonomous killer drones coming soon make SASR tame

By George Christensen

Ukraine has become the testing range for a new kind of warfare, one where drones do not just fly where they are told, but keep going when the signal is jammed, keep searching when the human link is cut, and in some cases reportedly select and strike targets without a live operator steering them.

That should stop you in your tracks.

Ukraine’s drone war has changed everything. In 2024, Ukraine reportedly acquired or produced nearly 2 million drones, including around 10,000 AI-enabled systems. These are not all high-end futuristic weapons either. Some are modified consumer drones. Some are more advanced military systems. Some are being developed or supplied by companies whose names are now becoming part of the new war economy.

The reason is simple enough. Jamming.

On the battlefield, electronic warfare can cut the link between the drone and its operator. A normal drone may lose control, fail its mission, or drop out of the sky. But an AI-enabled drone can keep going. It can navigate. It can recognise shapes. It can track movement. It can finish the job.

And that is where the moral floor starts to crack.

Reports from Ukraine have described a 2024 operation in which AI-powered drones were allegedly launched into a designated area without ongoing human control. A Ukrainian drone developer was quoted as saying, “We just launch it and we know everything will be dead.” Another reported quote was even colder: “Everything it sees will be killed.”

There are sentences that sound like they belong in a nightmare.

War is ugly. Nobody serious pretends otherwise. Men have been killing each other with spears, rifles, artillery, bombs and missiles since the beginning of civilisation. But even in the worst of it, there was usually a human hand in the chain. A soldier aimed. A pilot released. A commander gave an order. A person, a flesh-and-blood person with a conscience, a name and a soul, made the call.

Now the call may be buried in software.

A drone sees a truck. It sees heat. It sees movement. It sees a pattern in a database somewhere. Then it acts. Maybe it is right. Maybe it is wrong. Maybe the person below is a combatant. Maybe he is surrendering. Maybe he is wounded. Maybe he is a medic. Maybe he is just standing near the wrong vehicle at the wrong time.

Who answers for that?

The commander who deployed it? The programmer who trained the model? The company that sold it? The minister who signed the procurement papers? The operator who launched it and walked away? Or does everyone shrug and say the system behaved as expected within approved parameters?

That last possibility is the one that should frighten you most.

Because a world where nobody is responsible for killing is not a more advanced world. It is a darker one.

The Americans already have policy for this. Their Department of Defense Directive 3000.09, updated in January 2023, does not ban autonomous weapons. It says autonomous and semi-autonomous weapon systems must be designed to allow commanders and operators to exercise “appropriate levels of human judgment” over the use of force.

That sounds reassuring, until you think about what it actually means.

It does not necessarily mean a human presses the button before every strike. It can mean a human approves the mission area. A human approves the target category. A human signs off on the system. A human sets the rules, then the machine goes out and does what machines do.

That is not the same as human control. Not in the old sense. Not in the moral sense.

It is the language of bureaucracy wrapped around the transfer of life-and-death power from man to machine.

The Pentagon’s Replicator initiative points in the same direction. The whole idea is to field large numbers of cheap, expendable, autonomous and uncrewed systems, especially for a possible Indo-Pacific conflict. Cheap enough to lose. Smart enough to operate in contested environments. Numerous enough to overwhelm an enemy.

That is the future of war being built in front of us.

Not one pilot in one aircraft making one decision. Swarms. Loitering munitions. AI target recognition. Autonomous naval systems. Machines reacting to machines at speeds no human commander can properly follow until after the damage is done.

And Australia is not watching this from the cheap seats.

We are tied into the American defence machine through AUKUS, Pine Gap, intelligence sharing, joint exercises, procurement deals and all the quiet arrangements that most Australians only hear about after the decisions have already been made. If a major conflict breaks out in our region, especially in the Indo-Pacific, autonomous systems will not be some optional extra sitting in the corner. They will be central.

So where does Canberra stand? As usual, in the fog.

Our leaders talk about ethics. They talk about responsible use. They talk about international law, frameworks, guardrails and consultations. Penny Wong warned at the UN Security Council that life-and-death decisions must remain human responsibilities, especially as AI collides with nuclear risk and autonomous weapons.

Fine words. Necessary words.

But words are not a hard limit.

Words do not stop a drone once it has been launched. Words do not tell a defence contractor, “No, you may not build that.” Words do not tell an ally, “No, Australia will not be dragged into a machine-speed arms race without the people being told what is happening.”

That is the familiar Australian routine, isn’t it? Say the safe thing in public. Keep the real room for manoeuvre behind closed doors.

Even Pope Francis, speaking to the G7 in 2024, put the matter more plainly than many defence officials seem willing to: No machine should decide whether a human being lives or dies.

That is not weakness. That is civilisation.

You do not have to be a pacifist to understand it. You do not have to be soft on national defence. You do not have to pretend hostile regimes will play by the rules. Rogue states, terror proxies and criminal gangs are not going to abandon dangerous technology because Australia writes a nicely worded submission to a UN process.

We still need strength. We still need deterrence. We still need to defend ourselves.

But strength without moral limits becomes something else.

Once you allow machines to make lethal decisions over human beings, you have changed more than warfare. You have changed the value of human life. You have said, in effect, that a person made in the image of God can be reduced to a target signature and deleted by code.

That is where I draw the line.

And this is not happening in isolation. While militaries push AI into weapons, the civilian AI labs are warning that the technology itself may soon move beyond normal oversight. Anthropic has warned about the possibility of recursive self-improvement, where AI systems help build stronger successor systems, which then help build even stronger systems after that.

In plain English, the machines start helping build the next machines.

Anthropic has urged major AI labs to prepare a coordinated, verifiable pause if dangerous thresholds are reached. Not because the company is anti-technology, but because even people inside the industry can see the cliff edge coming.

That is the insanity of this moment. The people building frontier AI are warning that it may soon move faster than normal oversight, while defence departments are trying to plug the same technology into weapons that move faster than human judgment.

It all boils down to trust.

We are supposed to trust the same class of people who could not manage energy policy without punishing working families, could not tell the truth during COVID, could not secure their borders, and cannot even define basic biological reality without checking with activist lawyers.

Now they want us to trust them with machine-speed killing.

I do not. You should not either.

That does not mean every military use of AI is wrong. Of course not. AI can help analyse imagery. It can detect incoming threats. It can help defend soldiers. It can keep our people out of harm’s way. A drone used as a tool is one thing.

A drone given authority to choose and kill human targets is another thing entirely.

There must be a hard line. Not a vague line. Not a secret line. Not one of those slippery Canberra lines that changes meaning depending on which department is answering the question.

Governments should legislate that no fully autonomous weapon system may select and kill human beings without direct, meaningful human authorisation. The public should know what our defence policy is. Legislators should know what is being bought, built and integrated. Our soldiers should never be put in the position of outsourcing their conscience to an algorithm.

And no, this cannot all be hidden behind the usual words: classified, sensitive, operational, alliance obligations.

I am tired of that. A lot of Australians are tired of it.

When the government wants our sons and daughters to fight, it asks for trust. When it wants our money, it demands sacrifice. When it wants sweeping new powers, it tells us it is for our safety. Well, trust cuts both ways. If these technologies are going to redefine the battlefield, the people have a right to know where the moral boundaries are.

Because once this technology spreads, it will not stay neatly inside the hands of major militaries.

Cheap autonomous strike drones will end up everywhere. Smaller states will get them. Militias will get them. Terror groups will want them. Criminal networks will adapt them. Anyone with money, parts, code and hatred will look at this technology and see opportunity.

A drone does not need a runway. It does not need a uniformed pilot. It does not need a billion-dollar weapons platform. It can be small, cheap, disposable and deniable. Add AI recognition and a warhead, and suddenly the threat is not just on a battlefield overseas. It is near a public event, a church, a port, a power station, a police facility, a political rally, a farm gate or a suburban street.

People who dismiss this as alarmism have not been paying attention.

The international talks are moving, yes, but slowly. Too slowly. More than a hundred countries have been discussing possible rules on lethal autonomous weapons, but binding rules remain unlikely in the near term. That is the problem in a sentence. The diplomats are crawling while the engineers are sprinting.

And in the middle sits the public, mostly unaware, mostly excluded, expected to nod along whenever officials say the word “responsible”.

I do not accept that.

We need real red lines. No fully autonomous killing of human beings. No secret loopholes. No contractor trickery. No quiet alliance commitments that commit us to systems the people would never support if they understood them properly.

A machine can help a soldier. It can protect a soldier. It can gather information for a soldier.

But it must not replace the human soul in the decision to take a human life.

That is the line. And if our leaders will not draw it, ordinary Australians should force them to.

Because the age of autonomous weapons is not coming.

It has already started. And once machine killing becomes normal, it will be very hard to make the world remember why it ever horrified us in the first place.

Until next time, God bless you, your family and nation.