Tinder tests AI to auto-select photos from Camera Roll

Tinder is experimenting with a new feature using artificial intelligence to suggest to people on the dating app that they take safety precautions during the coronavirus outbreak, including washing their hands and “maintaining a social distance.” That’s the idea at least, according to Mark Irwin, Bumble’s director of technology. It’s a simple but ambitious request not only to understand personality and interests enough to recommend more compatible (and less creepy) matches across its app but also to deliver the types of inclusive conversations that keep women engaged.

What Chemistry does and where the feature is live

Chemistry, a dating site that is part of Match.com, is designed for people who are seeking help in getting to know someone online before meeting them for the first time. What is it offering? Chemistry enables its users to create detailed profiles using personal photos and interactive prompts.

If a user agrees, the system analyzes selected images for attributes and hobbies to use as match suggestions or profile display. The feature is currently in testing in New Zealand and Australia, where management has described it as a key pillar of Tinder’s 2026 product roadmap.

In practice, that could be surfacing outdoorsy matches if somebody’s Camera Roll is filled with trail selfies, or highlighting dog-friendly daters if there are obvious pet photos. Tinder already employs AI to help select profile pictures and to prod users with prompts like “Are you sure?” alerts before you send something offensive. Chemistry does take things further with its personalization, though, by taking into account first-party signals most heterosexual people don’t even realize are being entered in their profiles.

Privacy trade-offs and giving users control options

Requesting access to a Camera Roll raises obvious privacy red flags as well. Tinder asserts that the feature will be permission-based, though execution of it is the real test. Best practices here are:

• Clear opt-in flows for users
• On-device processing when feasible and practical
• Tight data minimization
• Mechanisms to easily exclude specific albums or images
• Transparency about retention practices and permission revocation

The shift is part of a larger trend. Meta recently experimented with applying AI to photos stored on a user’s device in order to suggest edits, another instance of apps hunting around for more intimate indicators from personal media. Regulators in Europe and the US have focused on consent and purpose limitation, and privacy-savvy users will demand granular controls and plain-language explanations of what is analyzed and why.

Why Tinder is investing so heavily in artificial intelligence

Tinder’s test is the latest indication that sustained pressure on growth has forced companies to rethink their strategies. Match Group has also seen nine straight quarters of declining paying subscribers at Tinder, as well as weaker spending among its younger users. On its latest investor call, the company estimated that Chemistry testing would decrease Tinder’s direct revenue by about $14 million during this quarter, as part of broader guidance of $865 million to $875 million, below consensus estimates around $884.2 million.

Recent performance underscores the challenge. Revenue from Tinder fell by about 3% year over year in the most recent quarter for which it reported, and paying users dropped by 7%. Overall, Match Group’s revenue increased 2 percent to $914.2 million, below consensus estimates, with earnings per share of 62 cents, or one cent short of expectations. Investing in AI capable of moving the needle on relevance and retention is a reasonable response, even if it hurts near-term results.

How this feature could change the Tinder user experience

If Chemistry operates as promised, people might encounter fewer generic matches and more substantial overlaps — say, surfacing climbers with other climbers, attendees of the same festivals or parties to one another, language learners with world travelers. The system might also tweak profiles by arranging photos that most accurately mirror a user’s interests and having them matched with prompts that elicit stronger conversations.

Tinder has been stacking up other engagement levers: dating modes for intent, double-date options, face verification and redesigned profiles that bring bios and prompts to the fore. AI is the connective tissue: it’s what can read signals from behavior, text and images to form a more accurate sense of each person’s vibe, then apply that understanding across discovery, safety and messaging.

Risks to watch and what Tinder users should consider

The calculus of risk and reward is based on trust. Some users may wonder if such an invasion of the Camera Roll is worth its enlargement. Even when it’s done on an opt-in basis, the optics can be tricky, and misclassification may — like seeing a friend’s baby photo instead of a signal about your personal life — have repercussions. Clear labeling, simple controls and the ability to review or delete inferences will matter.

For all who might be interested in using the feature: grant a barely acceptable set of permissions, check which photos Tinder has included and then revisit the settings after you have used it. Look for the company to issue technical notes on processing location and data handling; device-side analysis and short retention windows would be good signals.

The bottom line on Tinder’s new AI photo selection test

Chemistry signals the company’s bet that richer first-party signals — particularly out of photos — can revive matching quality and subscriber momentum. Upside: you get more relevant introductions and smoother profiles; Downside: more privacy sensitivity. If Tinder can provide measurable boosts in match acceptance, message response, and subscription conversion while respecting the strictest data security protections, it could be a meaningful advance for algorithmic matchmaking.