The Irish government (Cabinet) is due to consider a new digital and AI strategy this week, and the standout proposal is the Government’s intention to legislate to restrict social media access for those under 16 years old.
The political context matters. Ireland is heading into its EU Presidency (July–December 2026) and online safety is being positioned as a priority. The strategy is expected to reflect a familiar line: Ireland would prefer EU-wide rules, but will take national action if Europe moves too slowly.
Enforcement, Not Slogans.
The key question isn’t whether protecting children is important, it’s how the State can make any restriction meaningful.
Last year’s plan for a “digital wallet” age-verification pilot points to the enforcement challenge. An under-16 restriction is only as strong as the age assurance behind it. If access is still controlled by “enter your date of birth”, then the ban becomes more of a headline than a barrier.
But age checks raise a second concern, that of privacy. Any system must avoid becoming an online ID-by-stealth, or creating new data trails for children and families. If the public believes age verification means uploading identity documents for everyday apps, trust will evaporate quickly.
Algorithms are moving to centre stage.
One of the most significant elements of this debate is the focus on “recommender systems”, the algorithms that decide what users see next, i.e. instead of showing you posts in simple time order, the platform uses algorithms to predict what will keep you watching/scrolling/clicking and then automatically serves more of that.
The Joint Oireachtas Committee on Artificial Intelligence has urged that “recommender systems” should be off by default, and that platforms should be prevented from switching them on for children’s accounts. This goes beyond moderation and into product design, the mechanics that can drive compulsive use, extreme content pathways, and unhealthy comparison.
If Government wants to reduce harm, recommender controls may prove more effective than an age line on its own.
The wider EU backdrop.
The strategy also comes as European regulators intensify scrutiny of platforms and AI systems. It includes discussion about engaging the European Commission to ensure the EU AI Act’s prohibited practices remain fit for purpose as AI capability grows.
At the same time, Big Tech is lobbying hard. Meta has told Government it should prioritise efforts to scrap the planned EU Digital Fairness Act, which is expected to target addictive design and dark patterns. That alone signals where the next regulatory battles will be fought, not just content, but the way products are built.
Will it work?
Supporters say an under-16 restriction is a clear, protective line that reflects what many parents want. Critics, including CyberSafeKids CEO Mr Alex Cooney, argue a blanket ban could be porous, children will find workarounds and that it risks shifting responsibility onto families, rather than forcing platforms to reform.
Ultimately, Ireland will be judged on outcomes. If this is to be more than a headline, it needs three things, workable age assurance, credible privacy safeguards, and real obligations on platforms, especially around said “recommender systems” and addictive design.
Opinion: the ban headline isn’t enough, tackle harm by design.
An under-16 ban is an easy political sell. But it risks becoming a comforting story we tell ourselves while the underlying machinery remains untouched.
The platform problem isn’t mainly that teenagers exist online. It’s that many products are engineered to maximise attention, and the fastest way to do that is through emotional escalation, endless recommendations, and compulsive loops.
If Ireland wants a serious policy, it must do more than draw a line at 16. Age assurance has to be privacy-preserving, not a backdoor ID requirement. And the real test of ambition will be whether Government is prepared to confront “recommender systems“, the very engines that push users from one piece of content to the next.
A mature approach would target companies, not children, transparent design rules, meaningful enforcement, and algorithmic limits for minors. Otherwise, we’ll get a strong headline, and the same problems, just simply shifted around.
Leave a Reply