Aleksandr Zhadan was single and swiping Tinder in his Moscow apartment, looking for a girlfriend. Tired of sending out likes, messaging, and going on dates that went nowhere, he decided to program OpenAI’s GPT-2 into a dating assistant. His program talked with over 5,000 women, scheduling him over 100 dates. And it paid off: his Tinder AI bot led him to his wife.
Zhadan understands that dating apps can be hard and embarrassing. Once he’d heard about Chat GPT, using it to help break the ice was a clear move. His assistant sounded just like him, knew his interests, and scheduled dates for him. Little did his now-wife know she was talking with AI Aleksandr for the first few months of their ‘relationship’. In time, he transitioned the bot to his own messages, and soon, the bot was paused; the relationship went offline, and they moved in together in 2023.
Zhadan’s wife Karina learnt about the bot some months into their relationship, which didn’t anger her but did shock her. Initial chats on Tinder are surface-level enough, anyway. As Zhadan had worked hard to personalise the prompts, Karina saw the AI use as rational and practical, prioritising the real-life connection above all.
Aleksandr says that at some point during the project, the AI recommended he propose to Karina because she had expressed a desire to attend a wedding. ChatGPT thought she’d rather attend her own. He took the advice; she said yes. Their story is one end of the spectrum. Online dating can have tragic consequences.
And what does their story say about the impact of AI on relationships and how we start them? The lines between real human connection and robotic, inauthentic text machines have blurred to the point where you can find the love of your life. Is a relationship as real if it was created by ChatGPT?
Aleksandr and Karina’s story outlines one problem with online dating: how when we chat with anyone, we cannot be entirely sure that they typed them out. Your online crush’s images might be AI images generated, too, for all you know. Since ChatGPT launched in late 2022, the level of fraud and deceit on dating apps has soared. It was already a problem before generative text AI sites burst onto the digital scene. With AI getting more powerful and evidently more human, this is just the start.
A trusted world leader in bot management and account security, Arkose Labs fights fraud. The company provides support and technology to stop large-scale automated attacks and tackles dating app deceit. The US Federal Trade Commission (FTC) received 64,000 reports of romance scams in 2023, and losses to fraudulent online dating app users skyrocketed to $USD 1.14 billion. With median losses per person of $USD 2,000, it’s the highest reported loss for any form of imposter scam. The FTC comparatively received about 475,000 business imposter scams and about 228,000 government imposter scams.
In New Zealand, Netsafe received more than 600 romance scam reports in the last two years. Losses totalled $4.5 million. This echoes the 2024 Online Dating Norton Cyber Safety Insights Report from the consumer safety advocacy brand. Of the 1,007 adults surveyed, 28 per cent who had used a dating app said they’d been targeted by a dating scam, with 36 per cent of this 28 per cent falling victim. 24 per cent of Kiwis surveyed who said they’d used a dating app added they’d been catfished by someone.
Mark Gorrie, Managing Director APAC for Norton, says that while using AI as a dating coach to enhance your profile or even your photo when you’re navigating the world of online dating seems innocuous enough, not everyone uses AI technology to find true love. “While Kiwis think that online dating is relatively safe, it gets riskier and more complicated when used for less respectable reasons, like creating more convincing romance scams. People should stay vigilant for the common signs of a romance scam like avoiding video calls, having very few or low-quality images on their accounts or attempting to progress the relationship quickly.”
Perhaps surprisingly, the Norton survey’s results showed that over half of current online dating Kiwis are keen on using AI as their dating coach. Per the survey, these Kiwis spend an average of about four hours a week and more than $160 throughout their lifetime on dating apps and services. Of those who currently use an online dating app or service, 55 per cent are interested in using AI to write pick-up lines/conversation starters, 52 per cent for dating app profile development and 42 per cent for photo enhancement. Aleksandr is a pioneer of sorts, then.
Online dating has changed the way we meet new people outright. The survey also found that nearly a third (29 per cent) of Kiwi men prefer to meet a potential partner online, compared to only 14 per cent of women preferring to meet a potential partner virtually. Online daters ranked WhatsApp (53 per cent) and Instagram (51 per cent) as the safest dating apps.
Perceived safety is the key factor behind these findings. Beyond a strong password, being cautious with your personal information, and other generic advice, what can be reasonably said? The point of these apps is to fall for someone, which comes with slowly unravelling your inner self. Ultimately, we need to start prioritising real-life connections sooner and more honestly. Asking for a phone call or video chat is great for verifying the identity of the person you’re talking to. The issue then becomes whether that person is someone safe, period, no matter if the communication is in person or via an app.
Norton says that if you notice that the conversation you’re having seems off or the person isn’t answering your questions directly, odds are it’s an AI bot. Canned, predetermined responses and suspicious links (or links in general, because who is ever sending links on Tinder) are two red flags. Another tool to use is reverse image search. Check social media as well for blank or sparse profiles and no friend connections. In time, you will build Internet literacy and resilience towards fraudulent behaviour. In saying this, don’t rest on your laurels. Anyone – even the most confident child of computers – can be embroiled in a toxic romance scam or fall in love with an AI chatbot.
Dating apps collect more information about us than we realise. Information about your photos and device is collected from third parties, and metadata like when and where images were taken also paints a picture of your weekly behaviour. Using this data to guess more about you is common. The majority of
apps mention creating ‘inferences’ about you which is often code for targeting you with curated advertising to get more of your money. Dating apps sell this as improving the experience for you and working to help you find the one, but there is a fine line. Do these apps want you to find love, or do they want you to stay on their platform? As such, dating apps are designed to only ever be so helpful. The goal of integrating AI into online dating should be to foster genuine connections and enrich the user experience, not to keep users hooked on the platform indefinitely, but can we trust app companies to follow suit? When you consider this sentiment alongside AI chatbots and growing security concerns in our modern world, it can be a lot to reconcile. Maybe it’d be easier to stumble across the love of your life at your local.
It’s also worth noting that most apps are owned by one of two companies: Match Group and Bumble, which control over three-quarters of the US dating app market. Match Group owns a whopping 40 apps, including Tinder, OkCupid, Match, Hinge, Plenty of Fish, BlackPeopleMeet, OurTime, and dozens more dating apps. Spark Network owns Bumble, Badoo, Christian Mingle, Jdate, Elite Singles, and Zoosk. That’s a lot of personal data sharing.
The future of AI in online dating will see further integration and ethical concerns to match. AI can refine and personalise the dating experience while equally straying us further from authentic human warts-and-all connections. There’s also heightened responsibility put on users to be vigilant. Above all, the onus is on the apps themselves to handle user data carefully and ensure that AI doesn’t perpetuate biases or diminish the authenticity of meeting someone. As AI becomes more sophisticated, so too does the potential for its misuse. The argument for treating Tinder like LinkedIn is stronger than ever.
