By Tom Morgan, AI & Consumer Tech Analyst · Updated October 1, 2026 with Q2 2026 earnings filings and 2026 consumer surveys
Direct answer: In 2026, AI ranks what you see on dating apps and in online stores, and the evidence says it works best as a shortlist-maker you stay in charge of. People accept AI that ranks and suggests, resist AI that writes for them, and mostly refuse AI that acts for them. Two 2026 US surveys put the share of consumers who have let AI complete a purchase at 5% and 11%, and in one survey a majority of Gen Z and millennial daters said they’d be put off to learn a match used AI to chat with them.
Key numbers
- Hinge revenue grew 22% in Q2 2026; Tinder revenue fell 1% and Bumble revenue fell 15.2%.
- Match Group payers fell 6% to 13.3 million; Bumble paying users fell 16.4% to 3.2 million.
- 42% of US consumers used an AI tool to shop in the past month, but only 5% used a fully autonomous agent to place an order (NIQ).
- 76% of US adults in one survey are uncomfortable with AI completing purchases on their behalf (NMI).
- Nearly 60% of people who lost money to a romance scam in 2025 said it began on social media, and more than 40% of those who lost money to a social-media scam said it began with an ad for something they ordered (FTC).
The delegation ladder: one pattern, two markets
A swipe and a product click are the same event to the system behind them: a signal about what you’ll respond to. That is why dating and shopping now follow one pattern. Each product sits on a ladder of how much it does for you, and consumer comfort drops as it climbs.
The ladder is our editorial framework. The evidence column comes from the sources listed at the end, and those surveys use different samples and definitions, so read it as a pattern, not a measurement.
| Rung | Dating example | Shopping example | What the evidence says |
|---|---|---|---|
| 1. Ranks | Feed ordering, daily shortlists | “Recommended for you,” search results | Mostly invisible to you, yet experiments show it still shifts choices (see below) |
| 2. Suggests | Hinge Convo Starters and Prompt Feedback | Price comparison, deal codes | Well accepted: 63% would let AI insert discount codes at checkout (NMI), and 69% of Hinge daters are comfortable or neutral about AI helping plan dates |
| 3. Writes | AI-drafted replies and bios | AI-written product summaries | Resisted: 62% of Gen Z and 70% of millennial daters in one survey would be turned off by a match using AI in the talking stage (Hily) |
| 4. Acts | An assistant that runs the conversation | An agent that places the order | Rare and distrusted: 5% to 11% have done it, and 76% are uncomfortable with it (NIQ, NMI) |
Dating: the money shows what is actually working
Match Group and Bumble both reported Q2 2026 results in early August.
| Q2 2026 | Year over year | |
|---|---|---|
| Hinge direct revenue | $204 million | +22% |
| Hinge payers | 2.0 million | +17% |
| Tinder direct revenue | $457 million | −1% |
| Tinder payers | 8.5 million | −5% |
| Match Group total payers | 13.3 million | −6% |
| Bumble total revenue | $210.5 million | −15.2% |
| Bumble paying users | 3.2 million | −16.4% |
Sources: Match Group Q2 2026 release (SEC) for company totals and Hinge growth; Tinder and Hinge segment figures from Match’s Q2 earnings call; Bumble Q2 2026 results.
What the numbers say, and what they don’t:
- Hinge is the growth story. Match says it still expects Hinge to reach $1 billion in revenue in 2027, driven by product innovation, international expansion and monetization gains together. Hinge’s revenue grew 86% in its European expansion markets, which include France. The filing does not isolate AI as the cause.
- Tinder is slowing its decline, not reversing it. Match reports that Tinder’s daily active users fell 4% year over year, its best result in ten quarters, and that recommendation-algorithm updates are helping. That is a company claim, and it comes with a payer count that is still falling.
- Bumble is shrinking fastest, with revenue and paying users both down double digits.
- Revenue per payer is rising at Match (+6% to $21.13) and Bumble (+1.2% to $21.96) while the paying base shrinks. The filings don’t say whether that comes from pricing, mix or who is leaving.
Both leaders are also adding real-world and human layers. Tinder Events expanded to more cities (71% of eligible active 18–24-year-olds in the Los Angeles pilot engaged with the Events tab, per Match), and Hinge launched Friend’s Take in July, which brings friends and family into the process. Our read, not a sourced finding: the features gaining traction add a nudge or a human voice rather than replacing one.
What AI features do inside the apps
Hinge’s generative tools stop at rung 2 by design. Hinge says Convo Starters and Prompt Feedback do not provide pre-written language to copy and paste, only thought starters (Hinge, March 2026). The reasoning behind Convo Starters is Hinge’s own finding that 72% of daters are more inclined to consider someone when a like comes with a message, and that people who comment are twice as likely to arrange a date (TechCrunch).
What daters say about it
- High use, low tolerance for it being used on you. In the Hily T.R.U.T.H. report (1,559 US daters, reported by Fast Company), 82% of Gen Z and 87% of millennials said they use AI in dating, yet 62% of Gen Z and 70% of millennials said they’d be turned off if a match used it during the talking stage. Also, 53% of Gen Z and 66% of millennials said they’d feel less confident on a real-life date after using AI to communicate with a match. Hily is a dating app, so treat this as an industry survey.
- Definitions differ. Hinge’s “52% vibe-check messages” is a share of Gen Z daters who use AI, while Hily’s 82% to 87% counts any AI use among all surveyed daters. Don’t compare them directly.
- AI photos are the red line. 88% of Hinge daters say they’re uncomfortable with AI-generated profile photos (Hinge).
- Awareness is already high. A 2025 Norton study found 6 in 10 dating app users believe they’ve met at least one AI-written conversation (via Fast Company).
Shopping: heavy use for research, little for buying
The headline numbers depend on what counts as “using AI,” and 2026 surveys disagree for that reason.
| What was measured | Result | Source |
|---|---|---|
| Used at least one AI tool to shop in the past month | 42% | NIQ (US, about 500 respondents per wave, early 2026) |
| Used AI for product recommendations | 17% | NIQ |
| Engaged with an AI-powered shopping assistant | 10% | NIQ |
| Used fully autonomous agents to place orders | 5% | NIQ |
| Used AI to complete a transaction | 11% | NMI (1,000 US adults, Aug 2026) |
| Would give AI full control over a purchase | 10% | NMI |
| Would not trust AI to make purchases at all | 45% | NMI |
| Uncomfortable with AI completing purchases on their behalf | 76% | NMI |
| Say they must be able to review or override an AI decision | 70% | NMI |
| Have never uploaded payment details to an AI tool | 84% | NMI |
| Prefer a human to review transactions before an agent buys | 35% | RTB House via Retail Dive (1,800+ consumers, US, UK, Japan, France) |
The 5% and 11% figures look contradictory, but one counts autonomous agents placing orders and the other counts any AI-completed transaction. This is the same definition problem we covered in why AI agent adoption stats keep contradicting each other. Before trusting any adoption number, check the definition.
Safeguards change the answer. In the RTB House survey, 42% of US millennials said they’d let an AI agent buy within a $250 budget if they could return the purchase within seven days, versus about a third without that safeguard. NMI also found parents with children at home are nearly three times as likely to have completed a transaction using AI (19% vs. 7%), yet only 16% of parents would give AI full control.
Platform-reported usage tells a different story from surveys. Amazon’s CEO said on a recent earnings call that more than 350 million customers had used Alexa for Shopping in the past year (company-reported, via Retail Dive). Using an assistant to research is not the same as handing it your card, which is exactly the gap the surveys show.
For how platforms are building this, see our AI shopping trends analysis and AI shopping innovations.
Where both markets fail in the same three ways
1. The ranking shapes your choice, and you can’t see it
Experiments summarized in 2021 found that algorithms promoting some candidates over others influenced whom people chose to message or vote for, and that the effect depended on whether the promotion was explicit (such as a stated compatibility score) or covert (such as showing a photo more often) (ScienceDaily). The studies used fictitious candidates, so they show the mechanism, not the size of the effect on a real app. The practical lesson: the first screen is not neutral.
2. Your behavior becomes a profile
Your swipe rhythm and your cart history are both behavioral data. Our piece on how smart sensors turn movement into a behavioral profile explains how the same logic extends into the devices around you; the legal and compliance side is in AI and Privacy.
3. Scammers use both doors, and the same platforms
The FTC’s April 2026 analysis of 2025 reports ties the two markets together:
- Shopping scams were the most reported scam on social media. More than 40% of people who lost money to a scam on social media said it started when they ordered something they’d seen in an ad. Most said the goods never arrived; others got counterfeits.
- Romance scams also thrive there. Nearly 60% of people who reported losing money to a romance scam said it started on a social media platform, with $298 million in reported losses from those cases.
- The real totals are likely higher. The FTC notes that most scams go unreported and that it collected no reports during the 2025 government shutdown, so these are floors.
Visa’s security guidance adds a new angle for AI shopping: fraudsters are adapting search-optimization tactics to steer AI shopping agents toward scam sites that look trustworthy (Visa).
The shared signals are urgency, a too-good offer, and pressure to pay or send money in an unusual way.
The 7-rule playbook
- Stay on rungs 1 and 2. Let AI rank and suggest; write your own messages and make your own purchases.
- Never use AI-generated photos of yourself. It’s the least-accepted use in the Hinge data.
- Treat the first screen as a suggestion. Browse past the top results in both dating apps and stores.
- Require a confirmation step before any agent spends money, set a spending limit, and read the order before it pays. Seven in ten surveyed adults say they want to review or override an AI purchase decision.
- Know what you’re handing over. Screenshots, chats and purchase history are data; check storage and sharing terms before using any add-on tool, and don’t upload payment details to an AI tool unless you understand the permissions.
- Check before you pay or send money. Following the FTC’s advice, search a store’s name plus “scam” or “complaint” before buying from an ad, and treat any request for money from someone you haven’t met as a stop sign. Report scams at ReportFraud.ftc.gov.
- Verify “personalized” once. Compare one recommendation against a source the algorithm doesn’t control.
Frequently asked questions
Which dating app is growing in 2026? Hinge. In Q2 2026 its revenue rose 22% and its payers rose 17% to 2.0 million, while Tinder’s payers fell 5% and Bumble’s paying users fell 16.4%.
Is it okay to use AI to write dating messages? It’s common, but the surveyed reaction to receiving them is negative: 62% of Gen Z and 70% of millennial daters in one survey said they’d be turned off. Hinge’s own advice is to use AI as a second opinion and keep the message in your own voice.
Do AI shopping agents really buy things for people? Rarely. Two 2026 US surveys put it at 5% (autonomous agents) and 11% (any AI-completed transaction), and 76% of US adults in one survey are uncomfortable with AI completing purchases for them.
Are AI recommendations neutral? No. Recommendation systems are built to raise engagement or conversion, and experiments show ranking alone can shift choices. Use them to narrow options, then verify elsewhere.
How do I avoid scams that start on social media? Don’t send money to someone you haven’t met, check any unfamiliar store before you buy from an ad, and limit who can see your posts and contacts, as the FTC recommends.
Go deeper on AInvasion
- AI in Love & Dating and AI in Shopping: all coverage
- Best AI Dating Apps & Assistants: app-by-app review
- Do People Use AI to Flirt? · Rules of Dating · AI on a Dating App in 2026
- 25 AI Shopping Trends · AI in Shopping 2026
- AI Tools hub: free calculators and checklists, including shopping-safety tools
- Methodology · AI Glossary
Sources and confidence
| Claim | Source | Confidence | Checked |
|---|---|---|---|
| Match Group totals: revenue, payers, RPP, Tinder DAU −4%, Hinge +22%, $1B target, Friend’s Take | Match Group Q2 2026 release (SEC) | Established | Primary document read |
| Tinder and Hinge payer and revenue figures | Match Q2 2026 earnings call transcripts and slides | Probable | Secondary transcripts |
| Bumble Q2 2026 revenue, paying users, ARPPU | Bumble Q2 2026 filing and press release | Established | Primary excerpts |
| Hinge: 88% / 69% / 52%; no copy-paste language | Hinge newsroom, March 5, 2026 | Probable (company survey) | Read in full |
| Hinge: 72% more inclined, twice as likely to date | Hinge via TechCrunch (Dec 2025) | Probable | Excerpt |
| Hily: 82% / 87% use; 62% / 70% turned off; 53% / 66% less confident | Hily T.R.U.T.H. report via Fast Company (Nov 2025) | Probable (dating-app survey) | Read in full |
| 6 in 10 met an AI-written conversation | Norton 2025 via Fast Company | Probable (secondhand) | Read in full (relay) |
| NIQ: 42% / 17% / 10% / 5% | NIQ press release, May 5, 2026 | Probable (about 500 US consumers per wave) | Read in full |
| NMI: 11% / 10% / 45% / 76% / 70% / 84% / 63% / 19% vs 7% | NMI survey, 1,000 US adults, Aug 26, 2026 | Probable (payments-industry survey) | Read in full |
| RTB House: 35% prefer human review; $250 / 7-day return | RTB House survey via Retail Dive (Aug 2026) | Probable (multi-country, secondhand) | Read in full |
| 350 million customers used Alexa for Shopping | Amazon CEO, via Retail Dive | Probable (company-reported) | Read in full (relay) |
| Algorithms influence dating and voting choices in experiments | ScienceDaily summary (2021), fictitious candidates | Probable (dated, lab setting) | Excerpt |
| FTC: ~60% romance scams started on social media; $298M; >40% of social-media scam losers started with an ad | FTC Data Spotlight, April 2026 | Established | Primary page read |
| Fraudsters target agentic search results | Visa security guidance | Probable | Excerpt |
Confidence key: Established = confirmed in a primary filing or agency page · Probable = credible source, survey or secondhand relay · Speculative = not confirmed at the primary source.
Limits of this page. Most consumer figures come from company-sponsored surveys with different samples and definitions, so compare them as ranges, not as one number. Nearly all shopping surveys here are US-only. We did not test any dating assistant apps for this update and therefore list no prices or performance claims for them; see the full app review for tool details.