Beyond the catfish: building a community your dating app users can trust

Ask any dating platform what business they’re really in, and the honest answer isn’t matches or messages. It’s trust. Users hand you their photos, their conversations, and increasingly, their hearts, all on the assumption that the person on the other side of the screen is real. That assumption is now under attack. AI-generated deepfakes, synthetic identities, and fraud-as-a-service have turned what used to be a clumsy fake profile into something far harder to catch. The platforms that fail to adapt won’t just lose a few bad accounts. They’ll lose the one thing they can’t easily win back: their users’ confidence.

Listen to the Veriff Voices podcast, where experts like Ira and Geo unpack the threats and the solutions in their own words.

Produced by Any Other Business

What trust and safety leaders should do now

The threats are evolving faster than the naked eye can follow, and Veriff’s research confirms the human eye was never a reliable defense to begin with. The path forward is concrete:

• Stop fake profiles at registration. Verify that the person creating an account is real before they ever reach another user. Catching fraud at the front door is far cheaper than cleaning up the damage later.
• Detect deepfakes, don’t just check documents. Confirm that uploaded photos and live videos show genuine human faces, using detection built for the AI era, not the visual cues users wrongly trust.
• Build in layers, not single checks. Combine biometric, device, behavioral, and network signals so no single failure exposes your users.
• Don’t outsource vigilance to the user. With detection scores barely above chance and a high-risk segment that won’t verify anything, self-attestation is a liability. Make verification automated and technology-led by default.
• Treat safety as a feature. Communicate verification as a benefit your community can rely on, not an obstacle to endure.

The human eye can no longer tell real from fake

The single most important finding driving these recommendations is this: people cannot detect deepfakes reliably. Veriff’s 2026 research found that fake videos were frequently accepted as authentic while genuine ones were dismissed as fake, and for one female video pair, 70% of respondents misidentified the fake as real.

Deepfakes are different from every form of fraud that came before them. What makes them uniquely dangerous is that they don’t exploit a process or a system — they exploit human perception itself. Even people who work in this space professionally, myself included, find it genuinely difficult to distinguish real from synthetic after reviewing large volumes of deepfake content. For the general population, the odds are worse.

For a long time, the video call was the one barrier that held. A fraudster could steal someone’s photos, but going live on camera meant the lie would unravel. That’s no longer true. Real-time deepfake masking now allows bad actors to convincingly impersonate almost anyone on a live video feed. The case that received the most attention involved a woman who was defrauded by someone posing as Brad Pitt, complete with a fabricated medical emergency and requests for money. It sounds far-fetched until you understand how persuasive the technology has become.

If your trust and safety strategy relies on users, or even moderators, catching fakes by eye, it has already failed.

The most dangerous users think they can’t be fooled

There’s a second finding that should worry every trust and safety team: the gap between confidence and ability. Around half of US users believe they can reliably spot manipulated media, yet their actual detection scores hover right around guessing. As Veriff’s Fraud Platform Lead Ira Bondar-Mucci puts it, “This confidence-competence gap creates a false sense of security that fraudsters and bad actors are primed to exploit. When people believe they can’t be fooled, they stop looking for the signs.”

Roughly 7% of users fall into what the report classifies as “high-risk”: people who detect poorly, rate their own abilities highly, and rarely or never verify suspicious content. For a dating platform, that’s a precisely defined soft target sitting inside your user base, ready to be exploited by the next convincing fake.

Compounding the problem, the strategies people rely on no longer work. US users still look for unnatural skin (53%), oddities in hair or teeth (52%), and unnatural movement in video (51%), cues that modern deepfakes now replicate with ease. Checking content more often doesn’t improve accuracy, which suggests users simply lack a structured way to verify anything. The signals people trust have quietly expired.

Why a single detection check isn’t enough

Faced with deepfakes, the instinct is to look for one perfect detection tool, a system that scans a video or an image and returns a verdict. That instinct is a trap. Jolly’s advice is to treat deepfakes not as an isolation problem but as a broader context problem. A single check, he explains, simply isn’t enough: fake detection is no longer just about scanning the video. You need wider context, including how the user behaved beforehand, where they accessed the session from, what devices they used, and the resolution, length, and quality of the image or video.
Veriff’s approach reflects this. Its systems assess more than 1,000 data points alongside the image or video to determine whether it’s a real selfie, a real document, or a deepfake, doing automatically what humans, most of the time, cannot. A deepfake might slip past one check, but it struggles against the combined weight of evidence. The principle matters more than any single product. A system designed to catch everything in one interaction creates a single point of failure, and fraudsters are very good at finding single points of failure. Layered, context-aware detection is what stands up to industrialized, AI-powered fraud.

Fraud is now a business, sold by subscription

The reason single checks fail is that the threat has industrialized. The lone scammer in a hoodie is a myth. What Veriff’s experts actually encounter looks unsettlingly like a company. Organized fraud operations run shifts. They use device farms, rooms full of phones submitting verifications in parallel. They buy stolen identities in bulk from the dark web, rotate VPNs to disguise their location, and spoof devices to avoid detection.

Most concerning of all is the rise of fraud-as-a-service. Ready-made fraud toolkits are now sold on subscription, complete with pre-built fake document templates, real-time deepfake masking software, and step-by-step guides for bypassing specific verification systems. Some even come with live customer support over Telegram. “The barrier and the technical skill needed for those attacks have gone lower,” Bondar-Mucci says. “Fraud-as-a-service is democratizing fraud, which lets it scale at an amazingly fast pace. It’s a constant cat-and-mouse game.” The takeaway is uncomfortable but clear: the sophistication once reserved for a handful of expert criminals is now available to anyone willing to pay a monthly fee.

Why dating apps became a soft target

For years, getting onto a dating app required almost nothing. A photo. An email address. Maybe a phone number. As Bondar-Mucci puts it plainly: “The verification bar on most dating apps has historically been very low. That’s definitely not enough to stop someone from creating a completely fake profile.”

This wasn’t an oversight. It was a deliberate trade-off. In financial services, regulation forces companies to verify users. In dating, the biggest fear has always been friction, anything that might scare a potential user away during onboarding. So the industry under-invested in verification, precisely in the place where the human stakes are highest. The result is a sector that became a soft target. And bad actors noticed.

Book a demo

Your community deserves a platform built on trust.

Book a demo with our experts.

Book a demo

Catfishing was never really about the photo

It’s tempting to think of catfishing as a fake picture problem. It isn’t. The photo is just the door. The real damage happens behind it. As Bondar-Mucci explains, the crime isn’t really about the photo at all. It’s about manipulating a victim’s trust, building a relationship over weeks or months and then using it to extract money, personal information, or worse. Jolly, who has worked closely with dating platforms, traces the problem back to a single question: are you the person you claim to be? In dating, that question is deeply personal, and people’s insecurities can make them reluctant to show who they really are. The result is two distinct threats: catfishing, where someone hides their true self to find connection, and the romance scam, which is a deliberate exploitation of the ecosystem’s weaknesses.

Romance scams have already cost victims more than $1.3 billion in the US over five years, according to the FTC. But the figure only tells part of the story. The fuller cost is emotional: the betrayal, the distress, and the lingering sense that the platform itself couldn’t be trusted to keep its users safe. For a dating company, that’s an existential threat. Your reputation is your acquisition engine. The moment safety comes into question, existing users leave, and prospective ones never sign up.

Concern is high, but so is misplaced trust

Dating app users are not indifferent to this. Veriff’s research shows that nearly 8 in 10 Americans are concerned about deepfake-driven personal fraud and impersonation, with similar levels of worry about misinformation and the erosion of general trust online. That concern is well-founded, and for dating platforms, it’s a clear signal that safety is now part of the product experience users are evaluating.

But there’s a catch: Americans are also more likely than users in other markets to expect platforms and digital services to manage AI-generated content on their behalf. That creates a gap between concern and behavior: people recognize the risk, yet still rely on platforms to catch it, which can reduce personal vigilance at the worst possible time. As Bondar-Mucci puts it, the solution “isn’t to reassure customers, it’s to earn that trust through action.” For dating platforms, that means implementing verification that confirms a real person in real time and detects synthetic media at the moment of interaction, instead of expecting users to identify fakes on their own.

Verification as a safety feature, not a compliance box

The encouraging news is that the industry is changing. Dating platforms are beginning to treat identity verification not as a reluctant compliance formality, but as a genuine safety feature that users value. Bondar-Mucci notes that platforms in this space are starting to treat identity verification not just as a compliance checkmark, but as a real safety measure, going beyond document checks to confirm that profile pictures show real human faces, not deepfakes. Jolly sees the same momentum. The dating industry, he explains, is moving in a direction that focuses on trust and safety as a long-term priority, investing more not only in identity document verification but also in understanding whether each picture a user uploads is a deepfake, a positive shift he’s observed since last year.

This reframing changes everything. Verification stops being friction to minimize and becomes a promise to your community: the people here are real. Age validation extends that promise further, helping keep younger users off platforms they shouldn’t be on. Done well, verification doesn’t repel users. It reassures them. Trust, after all, is the product.

Build a platform your users can believe in

Catfishing was the warning. AI-generated deepfakes and fraud-as-a-service are the reality. And with detection accuracy barely above a coin flip, the platforms that thrive will be the ones that decide, deliberately, to make trust their defining feature, rather than asking users to fight this battle unarmed. Veriff helps dating platforms do exactly that, stopping fake profiles at registration, detecting deepfakes with 1,000+ data points of context, and giving users the confidence that the person they’re talking to is genuinely real.

Your community deserves a platform built on trust. Book a demo and let’s build it together.