Not all dating app AI does the same job. One predicts who you'll swipe on; the other checks whether the person you're talking to is real. Here's why Agilis only builds the second kind.
Pexels
When people hear that a dating app uses AI, they tend to picture one of two things: an algorithm that decides who you should meet, or a system that checks whether the people you're meeting are real. These are not the same thing. They don't serve the same purpose. And confusing them, or treating them as interchangeable, is one of the most consequential mistakes the dating industry is currently making.
At Agilis, we use AI in two places, and neither of them decides who you meet. The first is verification: checking that the person behind a profile photo is a real person. The second is safety: flagging messages that suggest harmful behaviour, so our team can act on them. What we don't do is hand your matches to an algorithm. Who appears in front of you isn't the output of a model trained on your swiping. That was a deliberate choice rather than a technical limitation, and understanding why means understanding what these two types of AI actually do.
Matching AI is the engine behind most modern dating apps. It observes your behaviour: which profiles you swipe on, how long you spend looking at them, which conversations you start, which ones you let die. It then uses that data to predict who you're most likely to connect with.
One major platform reported a 15% increase in matches since its AI discovery layer launched in early 2025, learning from conversational signals rather than swipe data alone. Across the industry, these systems factor in location, shared interests, activity patterns and behavioural history to rank which profiles appear in your feed. Source: SwipeStats, May 2026
On paper, this sounds helpful. In practice, it creates a system where the app is constantly optimising for engagement, not for connection. More matches means more time in the app. More time in the app means more subscribers. The incentive structure of matching AI points toward keeping you swiping, not toward helping you find someone.
There is also a subtler problem. When an algorithm decides who you should meet, it operates on behavioural data you may not even be aware of. As one industry analysis put it, these systems learn the preferences you act on rather than the ones you told the app you had, and they act on that difference. That is a system making consequential decisions about your romantic life based on patterns you have not consciously chosen or endorsed. Source: SwipeStats, May 2026
Matching AI, at its core, is about optimisation. It optimises for the platform's metrics first and for your outcomes second.
Verification AI does something entirely different. It answers a single, binary question: is this person who they claim to be?
It is not making judgments about compatibility. It is not learning from your behaviour. It is not ranking profiles or shaping what you see. It is checking whether the face in a profile photo belongs to the person trying to use the account.
This matters because the fake profile problem in 2026 is significantly worse than it was even two years ago. AI tools can now generate unlimited unique faces that have never existed anywhere online: faces with no reverse image trail, no history, and no way to be traced through traditional verification methods. Source: Social Catfish, July 2026
The consequences are not abstract. Reported losses to romance scams in the United States rose 22% in 2025, with an average reported loss of $2,020 per person, according to the FTC's Consumer Sentinel Network. Source: FTC, May 2026
Barclays claim data shows victims of romance scams in the UK lost £7,000 each on average in 2025, and 66% of UK adults say AI tools are making online dating scams harder to detect, with 53% concerned about having their voice or image impersonated. Source: Barclays, February 2026
Experian identified AI-powered romance scams as one of its top five fraud trends for 2026, warning that AI bots can now sustain convincing conversations, build trust over time and manipulate victims with precision. Source: ABA Banking Journal, January 2026
Verification AI is designed to address this at source. At Agilis, our verification combines liveness detection with facial comparison, establishing that there is a real, present person behind a profile photo rather than a generated image. It makes it materially harder for fake profiles to establish themselves on the platform.
The difference between these two types of AI is not technical. It is ethical.
Matching AI operates on your data, shapes your experience and serves the platform's engagement goals. You don't see it working. You don't consent to it specifically. And its incentives don't necessarily align with yours.
Verification AI, implemented responsibly, does something closer to the opposite. It uses AI to protect you: to establish that the person on the other side of a conversation is a real human being, not a synthetic persona designed to extract your trust and your money.
At Agilis, verification is optional, because we think you should be in the driving seat when it comes to your own data. If you choose to verify, your photos are used for that purpose alone. If you choose not to, that is your call. We use AI to help keep you safe, not to decide who you should love.
That combination, consent-first, safety-focused and user-controlled, is what separates verification AI from the matching AI that most platforms have built their business models around.
There is one more dimension to this conversation worth naming directly. Most dating platforms that rely heavily on matching AI do so because their business model depends on keeping you on the app and keeping you paying. A platform that makes money from subscriptions has a financial incentive to optimise for engagement rather than for outcomes. Verification AI serves a different master: it serves the user.
At Agilis, we made the app completely free, every feature, no paywalls, no premium tiers, because we believe the incentive structure of subscription dating is fundamentally broken. If you'd like to understand more about that decision and how we make it work, we wrote about it here: Why We Made Agilis Completely Free, and How It Stays That Way.
The dating industry is at an inflection point. According to Sumsub's Identity Fraud Report 2025–2026, online media and dating together record the highest identity fraud rate of any sector, with 6.3% of identity checks in that category turning out to be fraudulent. In the UK, Barclays' 2026 research found that 56% of Gen Z singles are now prioritising meeting partners in person over extended app-based conversations, with AI-related trust concerns cited as the driving factor. Source: Barclays, February 2026
Users are not naive. They know something has gone wrong. They are responding by disengaging: deleting apps, meeting people in person, opting out of a system that feels increasingly unreliable.
The platforms that will matter in the next five years are not the ones with the most sophisticated matching algorithms. They are the ones that can answer the most basic question a user has when they open a dating app: is this person real?
Verification AI answers that question. Matching AI, however sophisticated, is optimised for a different goal entirely, and that distinction is what the next phase of dating technology needs to reckon with.
Focused on getting a free, honest dating app in front of the people who need it most.
You're now subscribed to our newsletter.
