I'll admit something I don't usually say out loud in this line of work: there are moments when I no longer fully understand what is happening, and these moments are increasing. A few months ago, I read a proposal in more than one place about one vote per household, rather than one vote per person, floated as serious policy. It wasn't shouted, and it sounded almost "reasonable," which is exactly what makes it dangerous.

This proposal made me think that what we're watching isn't simply reversing course in some countries. It's the deepest dispute over what democracy and equality mean in a generation, and who gets to define them. I was born in Spain, under a dictatorship. I was a child when democracy arrived and I grew up watching it be built, watching my country learn what it meant to have a voice after decades of not having one. I have not seen this pace of change globally, this crossing of lines once considered untouchable, since.

It's worth separating three things that keep getting collapsed into one. The first is the proper backlash: the political and social reaction against gender equality, fed by the sense that it has produced undeserved gains or threatens existing norms. It doesn't always respond to a change that already happened, and often it mobilizes against the fear that one might. The second is backsliding: the legal, budgetary or operational weakening of past commitments. Sometimes it's brutal and visible, like the exclusion of women from public life in Afghanistan. Sometimes it's quiet: a unit that loses its functions, a budget line that shrinks without anyone announcing it, an equality mechanism that still exists on paper but not in practice.

And then there's a third phenomenon, harder to name but more dangerous: normative contestation. It doesn't try to stop one policy. It tries to weaken the frameworks that made progress possible in the first place, such as reopening what the phrase "gender equality" means, questioning affirmative action, chipping away at the state's very obligation to correct structural inequality. It's a dispute over language that when it gains ground, shrinks mandates and budgets before anyone notices. When language is hollowed out, so is the ability to demand anything.

This isn't an academic distinction. It changes what we actually do about it. Backlash calls for better narrative and proving the benefits of equality. Backsliding calls for institutional vigilance. Normative contestation can only be disputed by defending the conceptual ground itself, revisiting concepts, eliminating jargon and it is a challenge doing that with the rigour it needs.

I work inside an organization that operates within states, not alongside them, so I see this from a particular angle: the one where you have to help countries translate principles into budgets, electoral laws and finance ministries. And from there, the lesson is an uncomfortable one. We've spent years investing in equality mechanisms that can now be cut, redefined or hollowed out with surprising ease precisely because they never fully entered the hard core of the state. They were left at the margins: an office, a plan, a focal point or a budget line that's easy to strike out. I always wondered who wanted to be a focal point? A point? It is self-explanatory. Meanwhile, the decisions that actually structure inequality, such as fiscal policy, sovereign debt, trade regulation, investment treaties, kept being made at a different table, with no gender lens, and nobody thought to ask why.

That, to me, is the most urgent unfinished business: taking equality out of the rooms where everyone's already convinced and into the rooms where the conversation hasn't started. Central banks, trade or economy ministries, debt negotiations. As long as those decisions keep getting made without this lens, any normative progress will remain one election away from reversal.

And there's a frontier at the centre of this year's debate at the UN General Assembly that still isn't being looked at through this lens: the governance of Artificial intelligence (AI). Digital transformation isn't a neutral process. The algorithmic systems already deciding who gets a loan, a job or a benefit are reproducing biases that aren't always visible on the surface, and the platforms redefining the world of work are doing so with less regulation than existed for last century's economy. Whoever regulates AI now is deciding who's inside and who's outside the next two decades of development. This debate needs a gender lens from the start, not as an afterthought.

All of this is happening alongside something telling: wherever gender equality is most fiercely resisted, democratic quality tends to erode at the same pace. That's not a coincidence. Last week, UNDP released a new analysis of Latin America and the Caribbean that confirms with data what's visible everywhere else: the gap between the legal recognition of equality and its actual exercise remains across regions, and it is an early warning sign of broader democratic fragility.

Which is why resisting isn't enough. Reaffirming equality as a non-negotiable principle, yes, but also stepping into the rooms where the future is being decided: the economy, technology, security. We need to show concrete results and prove that equality isn't a cultural luxury but a condition for any democratic system to function at all.

The question isn't only how to defend what's already been won. It's what institutions, what coalitions and what language we need for gender equality to become, once again, a force that determines our future rather than one that merely resists its erosion.