In 2013 the Dutch climate foundation Urgenda, with 886 citizens, sued its own government. Urgenda argued the State knew climate change would seriously harm its people, yet had weakened its ambition from a planned 30% emissions cut to 14–17%, breaching the duty of care it owed them.
The Hague District Court agreed in June 2015, ordering a cut of at least 25% below 1990 levels by the end of 2020. The Court of Appeal upheld this in October 2018 on Articles 2 and 8 of the European Convention on Human Rights, the rights to life and to private and family life. The Supreme Court confirmed this on 20 December 2019.
The State argued that courts should not make climate policy, that the Dutch emissions share is tiny, that causation is uncertain, and that enough was already being done. The courts rejected each. The most influential holding, in my view, was that a state cannot deflect its share of the duty onto others, because “no reduction is negligible” (paras 5.7.8–5.7.9). I single this out because the small share argument has stalled climate politics for years. On separation of powers the line was careful: the court did not design policy but checked that the State met a legal minimum, leaving it free to choose the means.
The order was met: Statistics Netherlands confirmed 2020 emissions were 25.5% below 1990, helped by coal closures and Covid-19. Urgenda became the template for rights-based climate litigation in Germany, Ireland, France and at the European Court of Human Rights. The enduring outcome is the principle, not the percentage: where a state knows of a serious and foreseeable risk to life, human rights law obliges it to take reasonable measures against it.
I chose this case because it cuts through something political climate debate never resolves. Responsibility circulates endlessly between consumers, markets and states, and each can plausibly claim the problem, or the power to act, is not theirs alone. A court cannot be answered that way: it fixes the obligation on one party and asks only whether a legal minimum was met. That is litigation working as pressure where negotiation and parliamentary politics stall and lose valuable time.
The reasoning left me with a question. If a state must act on serious, foreseeable, scientifically evidenced risks to life, does that logic reach frontier AI, now widely described as posing catastrophic risk? Urgenda rested on settled scientific consensus and an agreed numerical benchmark. AI risk has neither yet, a limit of evidence rather than of principle. What other divisive future threats could it be applied to?
This was among my first attempts at researching a case in this field. I began broadly, briefing myself on the facts, then followed the case in sequence: why it was brought, what it was argued on, why the defences failed, what was ordered, and what followed. I worked from interviews and journalism through scientific sources to the judgments themselves, checking what I had read second-hand against what the courts actually said.
