When we started Slide, the dating app built by our team at Spotier, I thought I knew enough about safety. I had written about the trust problem in dating, and we had thought of verified profiles, Video Reels, Poke and Instant Chat as tools that bring people closer to reality. We had a list, an architecture and the belief that if you start with the right principles, the rest takes care of itself.
It does not. I have already written about what we are building with Slide and about how to fight fake profiles. Here I want to go a step further and put down, honestly and without triumphalism, what I have learned since about safety in a product where people put on the line not just their time, but their trust, their emotions and sometimes their physical safety.
Safety is a process, not a feature
The first mistake almost every founder makes, and I was no exception, is treating safety as a box to tick on the roadmap. You add verification, a report button, blocking, and move on. Except that bad actors do not read your roadmap. They test, adapt and look for exactly the spot you considered solved.
Safety has more in common with accounting than with a button: you do it every day, someone owns it, you revisit it whenever something changes, and you do not cut it when the budget gets tight. If nobody on the team clearly owns it, nobody does.
Verification reduces fakes, not bad behaviour
Verified profiles make a real difference. When you know the person in the photo actually exists, most of the crude fraud disappears. But verification only answers the question “is this a real person?”. It does not answer “is this a person acting in good faith?”.
A real person can be pushy, aggressive, send messages nobody asked for, or lie about their intentions. That is why verification is only the first layer. After it come community rules, moderation, reporting and consequences. A platform that boasts only about its verification badge gives users a greater sense of safety than is justified, and that can actually be dangerous.
Friction versus growth: the choice that hurts
This is the tension every app founder feels in their gut. Every safety step adds friction. One more step at sign-up, a video check, a confirmation before the first message: each one means some people give up. The sign-up chart looks less impressive, and investors ask why.
My answer is simple: it is worth it. Someone who quits because they are asked to show they are real is, in most cases, not the user you truly lose. The users you truly lose are the ones who have a bad experience and leave telling all their friends. The art is to put friction exactly where it stops abuse and remove it from everywhere else.
Moderation: people, tools and clear rules
Algorithms are good at volume: they catch patterns, flag suspicious accounts and filter obviously problematic messages. But context is something people understand. A joke between two adults and a threat can use the same words. Good moderation combines automated tools with people who make the final call in sensitive cases.
And those people need clear rules. If community guidelines are vague, decisions will be inconsistent, and users will feel it. Then there is speed: a harassment report handled a week later is, in practice, an ignored report.
Reporting must be easy and must lead somewhere
I have come to see that a report button buried three menus deep is almost the same as having none. Reporting should be one gesture away, right from the chat or the profile. Just as important is what happens next. If people report and never hear anything back, they stop reporting. A simple message such as “we received your report and took action” builds more trust than any campaign.
Interestingly, European law points in exactly the same direction. The Digital Services Act, which has applied to all online platforms since February 2024, broadly requires an easy way to flag illegal content, reasons for moderation decisions, a way to appeal them and transparency reports. For a team that wanted to do things properly anyway, the DSA is not a burden but a useful checklist.
Protecting women and vulnerable users
In a dating app, risk is not evenly distributed. Women typically receive far more unsolicited messages and more aggressive behaviour. If you design the product for the “average user”, you leave exposed precisely the people who need protection most.
In practice, that means control: who can contact you, when and how. It means instant blocking, with no explanation required and no notification to the other person. And it means never revealing anyone’s exact location. An approximate distance is enough to find people nearby; a precise location can become a physical safety problem.
Scams evolve, so verification has to evolve too
Classic romance scams, with sob stories and requests for money, have not gone away. They have been joined by investment scams, especially crypto ones, where the “partner” does not ask for money but teaches you how to “invest” on a platform they control. It looks like a relationship, but it is a patient fraud scheme.
Then came generative AI. AI-generated photos look more and more convincing, and fake video, which I wrote about in the context of deepfakes and elections, is now within anyone’s reach. A check that was solid two years ago may not be enough today. My conclusion: verification is never a finished project, but a race in which you have to stay at least one step ahead of those trying to fool it.
Sensitive data, and being open when you get it wrong
A dating app knows extremely personal things about its users: what they are looking for, who they talk to, sometimes their sexual orientation or where they are. That data should be collected sparingly, kept as briefly as possible and protected seriously. The rule I follow: if you do not need a piece of information for the product to work, do not ask for it. I applied the same logic when I wrote about building a responsible AI companion.
And since nobody is perfect: when you get something wrong, say so. A wrong moderation decision, a feature that created an unexpected problem, a clumsy piece of communication. Users forgive an admitted mistake far more easily than one that was hidden and discovered later.
Practical tips for anyone using dating apps
Whatever app you use, a few simple rules make a big difference:
- Never send money, gift cards or crypto to someone you have not met, however convincing the story.
- Be suspicious of anyone who brings up an “investment opportunity”, especially after a few weeks of chatting.
- Favour verified profiles and ask for a video call before meeting.
- Meet for the first time in a public place, with your own transport, not at anyone’s home.
- Tell a friend where you are going and who you are meeting.
- Do not share your address, workplace or financial details in early conversations.
- Block and report without hesitation. You do not owe anyone an explanation.
Trust is earned every day
If I had to sum up what Slide has taught me in one sentence: safety is not something you build once, but a promise you keep every day. It costs sign-ups, time and money. But in an industry built on trust, it is the only asset that matters in the long run.
What makes you trust a dating app? And what has ever made you delete one?