Lovable Apps Exposed Passports and Payment Data, Study Finds
German security firm Crissy Field scanned 159,041 Lovable projects and found sites leaking passport details, payment data and over 200,000 chat messages without any login.

Apps built with the AI platform Lovable have exposed sensitive user data to anyone who knew where to look, according to new research by the German cybersecurity firm Crissy Field. The exposed material included passport details, payment information, personal records and more than 200,000 chat messages, all reachable without logging in.
Lovable lets users create apps and websites simply by describing what they want, with no advanced programming skills required. The investigation suggests that making development easier does not automatically protect the data those apps handle.
What the researchers found
Crissy Field examined about 159,041 projects built with Lovable. Of those, 34,545 used the Supabase database service, which is commonly paired with the platform. In 24,664 cases, websites sent database records as soon as a page loaded, without asking visitors to sign in.
That does not mean every one of those sites leaked private data, since some databases hold information meant to be public. Still, the researchers ran additional read-only requests and found records containing email addresses on 7,403 sites.
In one of the most serious cases, a service exposed more than 1,000 identity records with passport numbers, nationalities and expiry dates. Other apps left customer payment details, phone numbers, postal addresses and ticket codes open. Contracts, invoices, personal photo albums and internal WhatsApp conversations were also visible. Such data could be used for phishing, identity theft or ticket fraud, though the investigation could not show that anyone actually misused it.
Responses and caveats
Lovable told the business daily Handelsblatt that the problem is not limited to its platform but widespread across the app-building industry. Its terms of use make customers responsible for the apps they create and for how those apps handle data.
Crissy Field cautioned against treating the results as a confirmed leak rate for all Lovable apps. The sample was gathered through Common Crawl and does not cover every app on the platform, and the researchers did not download entire databases in every case.
The findings add to growing concern about AI-assisted coding tools: they lower the barrier to building software, but securing databases and managing access permissions still demands careful attention from whoever publishes the app.
Apps worth a look
Tap the right tiles, add them up, hit the target. Easy to learn, hard to master. Free, for all ages.
Talk to an AI conversation partner; every sentence comes back corrected and explained in your language. 9 languages, A1 to C2.
Open EDIFACT and X12 messages, see every segment explained, convert to JSON/XML, export PDF and Excel. Runs locally, no upload.
More from this day
See all →
Cloudflare Acquires Deno, and the Runtime Is to Be Wound Down

OpenAI: ChatGPT Web Search Volume Hits 20% of Google's

Freshworks Posts $18.3 Million Q3 Profit as Revenue Climbs 14%

IndiaAI Mission's Three Indigenous AI Models Target Offline and Secure Use
More in: Security
See all →
Avast Launches Free Unlimited VPN With No Ads in France

Hacked fridges: supermarket cooling systems hit across Europe

Sega Fave Hit by Breach: Arcade Online Services Suspended
