16 November 2026 · 5 min read · BlitzHash · AI & Vibe Coding

Lovable data exposure research: what it found, what it didn't, and a 10-minute check

Short answer: in November 2026, the German security firm Crissy Field reported that many Lovable apps built on Supabase let a signed-out visitor read database records, including some highly sensitive ones. The headline "30%" figure is not a confirmed leak rate, and the researchers say so themselves. What matters for you is simpler: if your Lovable app stores customer data in Supabase, check it today, because the platform changes shipped so far don't close tables that are already open. You can start with our free leak check on an app you own, or the free self-check, in about ten minutes.

This post is for founders and small teams who built with Lovable (or Bolt, v0 or Cursor) on top of Supabase and aren't sure whether their database rules are doing their job.

What the research actually found

Crissy Field published the work with the German business paper Handelsblatt, which reported it on November 2, 2026. The updated English report, Lovable: Many Apps Built on the Platform Expose Sensitive Data, reached the Hacker News front page on November 14 and is still being discussed.

Here is what they report, in their own numbers:

  • They found 159,041 Lovable projects in the Common Crawl web archive (December 2025 to September 2026). 34,545 of them used Supabase.
  • 24,664 of those sites sent database records as soon as the page loaded, with no login. The researchers point out that this alone is not a security problem, because some data is meant to be public, like a menu or an event list.
  • With further read-only requests using the same public key, they found records containing email addresses in 10,560 tables across 7,403 of those 24,664 sites, about 30% of the group they examined more closely.
  • Cases they inspected included a forum with more than 80,000 profiles and more than 200,000 chat messages, ticket and food-order records with payment information, and more than 1,000 records with passport details.

What the research did not find

This is the part most social posts skip, so it is worth reading slowly:

  • The 30% figure is not a confirmed data leak rate. Email addresses can be public contact details or test data, and the method misses personal records that have no email field. The researchers say these effects partly cancel out and that they could not measure by how much.
  • It is not a rate for all Lovable apps. The Common Crawl sample is not representative of every Lovable app.
  • They could not establish whether any of the exposed data was misused.
  • Nobody broke into Lovable's own servers. The records were readable through the public key that a Supabase front end is designed to ship, because the access rules on those tables were missing or too loose.

What Lovable and Supabase have said and changed

Lovable's Head of Security replied on the Hacker News thread on November 14. He thanked the researchers and said the root cause in these apps was insufficient Row-Level Security, which Lovable's scanning is built to catch. He also said Lovable scans every app before publish, offers a free deeper scan, a Fix button and auto-fix, and lets you block publishing when there are critical findings. In a later reply he said his advice for now is to go and click the Fix button, and that more releases are coming to raise the security baseline.

Two platform changes are relevant, and neither repairs an app that is already live:

  • Lovable's November 12 changelog says projects on its built-in backend with unresolved critical findings now get asked to review them before publishing. You can still choose to publish anyway.
  • Supabase's breaking change on Data API exposure means new tables in the public schema are no longer reachable through the Data API until you grant access, applied to existing projects from October 30, 2026. Supabase is explicit that existing tables keep their current grants and stay reachable.

So if a table was open on October 29, it is still open today unless someone fixed it.

A 10-minute check on an app you own

Only test apps you own, and use test accounts, never real customers. Don't paste passwords or keys into forms or emails.

1. Run the free leak check. Paste your live app's address into our leak check. It reads the public page the way a stranger would and tells you what is already visible. It does not sign in, change data or keep the rows. 2. Run Lovable's own scans. Open the Security view in your project, run the deep scan (it is free, but it does not run on its own), and read every critical finding. Don't click "publish anyway" on something you don't understand. 3. Do the signed-out look. Open the published app, not the preview, in a private window while signed out. Public pages are fine. A customer list, an order, a private file or a message thread is not. 4. Do the two-account test. Sign in as test user A and note a record that should be private to A. Sign in as test user B and try to open it, including any direct link. If B can see it, the rules are not doing their job. The full version is in our two-account test. 5. Check Supabase's Security Advisor. In your Supabase dashboard, open the Security Advisor and look for tables without Row-Level Security, then read the policies on every table that holds personal data. Our 30-minute Supabase RLS walkthrough explains what "RLS is on but data is still open" looks like.

If any step shows something private, stop testing on real data and get a person to look. A scanner flags patterns. It can't know your business rules, like which customer should see which order.

If you want a second pair of eyes

We review AI-built apps for exactly this: Supabase access rules, exposed keys, authentication, and what a stranger can reach on the live app. See the Lovable app security audit and Supabase RLS check pages, or book a free 15-minute launch review. We've worked with 45 clients across 50 projects, and you keep your code, accounts and data.

Start with the free self-check. It takes a few minutes and tells you where to look first.

Sources

Related posts

Next step

Book a free 15-min launch review.

A short video call. We look at your AI-built app, flag the biggest risks in security, launch readiness, speed and cost, and you leave with next steps. No obligation.

Every price is a custom quote after we look at the app. No work starts until you accept the proposal.