Prohibition of Lethal Autonomous Weapons Systems Bill 2026: Second Stage
Seanad Éireann — 2026-06-17
============================================================

Alice-Mary Higgins (IND), National University of Ireland

I welcome the Minister of State to the House. I also thank the experts who have worked with me on this legislation, and Stephen Reid from my own office for his work on its development.
One of the moments when I was proudest of Ireland was being there when Ireland hosted the negotiations which led to a global ban on cluster munitions. Cluster munitions were land mines from the sky, weapons which by their very nature hurt unknown targets at unknown times in the future, were unpredictable in their effect, and were outside the realms of international humanitarian law because their consequences were unknown. What we did know was that their consequences were very often at the expense of civilians and children. Ireland led the discussions that led to the global ban on these horrendous weapons. That is why when people talk about Ireland stepping up and playing our role, I always think that Ireland plays its role, and the role it has played has been in peace and disarmament. That is our outsized role, and it is the greatest contribution a country makes in terms of reduction of casualties and of death.
We are now at a moment when not only are we seeing a very worrying rollback, with countries leaving the land mine and cluster munition treaties and returning to the use of these unconscionable weapons, but we are also seeing horrendous new weapons which are dealing death at a scale and with mechanisms which were unimaginable in the past. Lethal autonomous weapons have been described by the Secretary-General of the United Nations, António Guterres, as morally repugnant. Pope Leo, in his encyclical, spoke about the horror of entrusting an algorithm with the power to select who is worthy or not without bearing responsibility for that judgment, and about the damage that does to human possibilities and political responsibility. He highlighted that artificial intelligence does not remove the intrinsic inhumanity of the conflict. In fact, it brings about conflict more quickly, renders it more impersonal, lowers the threshold for resorting to violence, transforms defence into threat prediction and reduces victims to data.
The legislation I am proposing now comes in a context. I will dive a little bit more into autonomous weapons in a moment. Not only are we seeing horrific weapons being used with consequence around the world, but we are seeing a turbo-powered development of such weapons. As they are coming about, they are being experimented with. They are being used in Iraq and Iran, and on the population of Gaza. They are being trialled and tested on the most vulnerable people in the world. This is being done in the context of more money being pushed towards defence than we have seen in almost 60 years, and more money being poured into AI. The two drivers - the AI industry and the defence industry - are having vast amounts of money thrown at them in the kind of "move fast and break things" epic we have seen in tech in the past. In this instance, what is getting broken is humanity and people, and indeed our international humanitarian law and those ideas that there are standards even within war.
As a sense of how these are used, I could use quotes about Project Maven, which was initially developed by the Pentagon and is now available in all contexts. It can identify approximately 1,000 targeting decisions in an hour. It can choose and dismiss targets in the battlefield. One director, the vice admiral, has said that America wants to use it for everything, not just targeting. I know my colleague will speak more on Israel, where we have seen the Gospel - a machine-learning algorithm - used to produce potential targets so fast that Israeli officers described it as a "mass-assassination factory". A similar system, Lavender, at one point identified 37,000 potential targets. When those in Israeli intelligence have spoken about how it felt about to use these weapons, they have downplayed the role of human oversight. They have said that they would invest 20 seconds for every target at this stage and do dozens of them every day. They have spoken of how they had zero added value as humans, apart from as a stamp of approval. I know my colleague will speak of other systems, including Hello Daddy, where by their nature, the targets effectively are children.
These weapons are being used and trialled, and there is a giddiness in their adoption. This is happening with more money than we can imagine coming from two directions, yet there is a massive lacuna in terms of legislation, accountability and structures. The EU AI Act, which maybe gets spoken about in relation to other areas of AI regulation, with all of the concerns which come with the omnibus and others, explicitly does not deal with the area of military or defence or even dual-use, which is something Ireland needs to have a strong moral eye to, given the very large number of dual-use goods, for example, that we export to Israel and elsewhere. Those are explicitly excluded from the EU regulatory context. The national context has not taken real action. Internationally, we know there is a process that has taken far too long to come to the point of maybe looking to an adaptation of the Convention on Certain Conventional Weapons, even though we know the General Assembly of the UN has also said it will look otherwise for treaty action or for action on this. There is a gap in the governance at a time when the escalation and the funding are immense.
I want to touch on a couple of the key components in this legislation. Our Bill seeks to restrict the use, stockpiling, production, transfer and development of and investment in lethal autonomous weapon systems. It would effectively ban the Irish Government and indeed the Defence Forces from utilising this classification of weapon. It would curtail direct or indirect investment of public funds into this area and indeed the transit of weapons through this territory. We take our definition from the definition that I know the Government has looked at in its engagement with the UN process, which is the International Committee of the Red Cross, ICRC, definition, which looks at lethal autonomous weapons or any weapon with autonomy in its critical function, or any weapons system that can select and attack targets without human intervention. These are commonly known as killer robots. The killer robot campaign makes a number of very important points, which I will not have time to elaborate on in detail but I want to touch on.
One of the key issues here is digital dehumanisation. Technology should be empowering all members of society rather than reducing us. Used against people, it simply profile-pattern matches and processes human beings as data. Fundamentally, machines do not recognise people as people; they simply recognise them as a set of patterns or pieces of data to be processed. It is one of the most essential dehumanisations applied to the most vital decision, which is whether a human lives or dies. If this goes to the area of machine decision making, there are risks to other areas of our lives such as the question of exacerbating existing inequalities. What we know from many algorithmic areas and areas where AI has been applied is that it absorbs and amplifies prejudices and discriminations.
For example, I mentioned Maven. In certain contexts, its accuracy drops below 30%. In relation to policing, we know that AI measures that have targeted and strongly impacted minority groups, in particular, ethnic minorities.
Alienation has been shown to lower the threshold to war. One of the crucial aspects of the processes that are under way is definitions of meaningful human control. In my Bill, I make it clear that someone sending an instruction to what kind of targets they would like is not meaningful human control and does not satisfy that standard. There is a huge danger that we will move to the so-called bias, where 20 seconds are given to press the button. Crucially, machines cannot make those complex ethical choices. Under the law, military commanders have to judge the necessity and proportionality of each attack. They have to distinguish between civilians and military targets. That means not only understanding the weapon systems they are using but understanding the context in which they are used. That context and those consequences cannot be properly reflected if we move towards autonomous weapons.
I thank Professor Luke Moffett from Queen's University Belfast who has done great work on this issue and also on warification. These kinds of weapons cannot identify a boat on which people are waving a white flag or making radio communications that they are shipwrecked. In Gaza, drone operators could not identify World Central Kitchen vehicles because they were using infrared and the organisation's logo did not show up. Those who make an attack cannot necessarily see that a children's school bus is coming up right alongside their target because they are focused on their target. This is what happens when we lose human control. After the loss of control, as Pope Leo mentioned, comes the idea of political responsibility and accountability being lost. We lose track of who can be blamed, and even what happened because some of these systems operate with such a black box that there is not only a lack of understanding about what should be done but also a lack of understanding of what happened. In that regard, we need that understandability.
At this time of this arms race, I urge that Ireland show leadership, support this Bill and be a strong driver of the push, internationally and at national level, for a prohibition on these arms and weapons. I look forward to the Minister of State's response.

---

Source: Houses of the Oireachtas. Licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). The Official Report is revised after first publication; the fetch timestamp below identifies the version quoted.

Record URI: https://data.oireachtas.ie/akn/ie/debateRecord/seanad/2026-06-17/debate/main
Retrieved: 2026-08-14T04:55:04+00:00
Sitting date: 2026-06-17