Artificial Intelligence: Statements Seanad Éireann — 2026-01-28 ============================================================ Lynn Ruane (IND), University of Dublin I set this speech against the backdrop of a quote that underpins a lot of the work I do, especially in the evolving area of AI. It is from a book I read a few years ago by Virginia Eubanks called Automating Inequality. It states, "When automated decision-making tools are not built to explicitly dismantle structural inequities, their speed and scale intensify them". We know that around the world artificial intelligence is already shaping people’s lives in warfare, public services, policing, welfare, education and online. From the outset we must be clear about one thing: AI is not neutral. The way it is designed, deployed and governed reflects choices about power, priorities and values. These are not merely technical decisions but ethical and political ones. That is why AI must be grounded in human rights and equality, not treated solely as an economic or technological issue. A technology-first approach is a bad approach. What we need instead are human rights and equality by design, built in from the very beginning. AI and its consequences dilute responsibility and shift it away from accountability. Recent events involving Grok, the AI system integrated into the X platform, bring this debate out of the abstract and into the realm of real-world and ongoing harm. This is not simply a regulatory or ethical issue but a criminal law issue under existing Irish law. Ireland already has clear laws prohibiting the creation, publication and distribution of child sexual abuse material and non-consensual intimate images. Those laws apply regardless of whether the material is produced by a human or by an automated system. Where an AI system generates sexualised images of real people without consent and makes them available to others, publication and distribution are inseparable. In such cases platforms cannot credibly present themselves as passive intermediaries. They are operating the systems that produce and disseminate the material. Under Coco’s Law, it is a criminal offence for a person or a corporate body to publish or distribute non-consensual intimate images. In the case of Grok, publication and distribution occur at the moment of creation. Automation cannot serve as a shield against criminal responsibility. This is not a matter of waiting for the AI Act or future frameworks; this law already exists. The harm is foreseeable and the conduct unlawful, yet the response appears to rely on future-facing processes rather than the immediate use of existing enforcement powers. This matters particularly for poorer and marginalised communities, who consistently bear the brunt of harmful AI deployment. Across the world we see AI systems being used in welfare administration, housing allocation, credit scoring, immigration control and policing. Too often these systems automate suspicion and intensify surveillance. When these systems fail due to error or bias, it is those with the least power, time or resources who pay the price. This is how inequality becomes automated. We have also seen where AI has gone wrong globally. Examples include facial recognition systems that misidentify people of colour, predictive policing tools that reinforce racist and class-based patterns and automated content systems that amplify fear and harassment faster than regulators can respond. These are not glitches but the predictable outcomes of deploying powerful systems that take away our control and autonomy. We need to clearly determine who benefits from the proliferation of AI. Artists and musicians are having their creative work stolen and instrumentalised in AI applications. Workers are being replaced by AI systems. Amazon today announced 16,000 job losses and last year the company’s chief executive told staff that AI meant the company would operate with fewer workers. The result of this automation is less to pay in wages and more profit for Amazon. We must confront the even more uncomfortable reality that AI is increasingly being used as tool of war. AI-enabled decision support systems are already being deployed in military operations. In Gaza, such systems have reportedly been used to generate large-scale targeting lists with minimal human oversight. One system known as "Lavender" produced a list of tens of thousands of people marked for assassination. Another system referred to as "Where's Daddy?" was designed to track individuals at home at night explicitly to enable strikes when families were present. These were described as decision support tools, and not weapons, yet they directly facilitated lethal force at scale. Human rights law does not distinguish between harm caused by a missile and harm enabled by an algorithm and neither should our regulatory research or procurement frameworks. Returning to the domestic context, the Grok case exposes serious questions about enforcement and deterrence. Irish regulators already have the power to investigate, restrict and sanction harmful activities. If these powers are not being exercised, we must ask what threshold of harm is required to trigger fundamental rights in Ireland. Where AI systems generate and publish harmful material in a single automated process, accountability must follow. This leads to an unavoidable question. What real deterrent currently exists for AI providers that knowingly deploy systems with foreseeable abusive capabilities? We cannot wait for Europe to act. The EU omnibus bill is presented as simplification but in practice, it erodes core GDPR protections as AI systems expand surveillance, profiling and automated decision-making, thereby gutting privacy law and data rights. We must honestly address the enormous environmental and climate ramifications of the rapid increase in AI infrastructure. OpenAI's own researchers have shown an exponential increase in the energy required to run these models with computing requirements doubling every 3.4 months. In Ireland, data centres consume more energy than the domestic use of 3.25 million people in urban homes. We must reckon with the fact that the rapid increase in data centres and associated AI infrastructure is introducing an entirely new category of environmentally damaging activity not only in terms of the existential threat posed by climate change but also the pollution caused by these centres affecting our air, soil and rivers. As Ireland prepares to increase the share of our electricity grid dedicated to data centres, we must also engage with the associated financial risks of over-exposure to this sector in the Irish economy. Economists worldwide have raised concerns about the financial interdependence among these large companies and the need for continued growth. We must address our heightened exposure to this market. Finally, the public sector must lead by example. From procurement to deployment to monitoring, AI systems used in public services must be rights-compliant from the outset, rather than retrofitted after harm has occurred. Human rights and equality cannot be an afterthought. Procurement must be transparent, it must assess all risk - environmental, equity, discrimination and potential abuses - and it must be very clear on who is responsible for addressing these risks. --- 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/2026-01-28/debate/main Retrieved: 2026-08-14T04:54:33+00:00 Sitting date: 2026-01-28