Skip to content

Finding

Some database tables answer without anyone signing in

A request made with the browser key alone returned rows, which means the row rules on those tables let anonymous readers in.

Last updated September 20, 2026

Before launch · Needs a developer

Why AI-built apps get this

AI builders create tables in seconds and leave the access rules for later. The browser key is published in your app by design, so a table with no rule behind it answers anyone who copies that key. This is CVE-2025-48757 across 170 apps, and the Moltbook exposure, in one sentence.

What it affects

A request made with the browser key alone returned rows, which means the row rules on those tables let anonymous readers in.

What the evidence line looks like

table profiles answered the published browser key with 42 rows

Every finding on a result carries a line like this one: the request that was made and the answer that came back. You can run the same request yourself and get the same answer, which is the point.

The fix

Turn on row level rules for each table listed and write a policy scoped to the signed in owner. Tables returning rows: {tables} of {total} readable.

Prompt for your coding agent

These tables return rows to anyone with your public key: {tables}. Turn on row level security for each and add policies that only let the signed-in owner read their own rows.

A result gives you this prompt written for your builder, in its own dialect, and a batch prompt that carries every fix at once.

Incidents that involved this

  • Not verified

    First-quarter 2026 assessment of 200 apps

    183 of 200 vibe-coded apps, 91.5 percent, contained at least one vulnerability traceable to AI hallucination or missing security context.

  • Observed

    Moltbook records exposure

    An app exposed 4.75 million records, including 1.5 million API tokens and 35,000 email addresses.

  • Not verified

    Tea app, first breach

    An unprotected storage instance exposed tens of thousands of user images, including identity documents.

  • Not verified

    Tea app, second breach

    Three days after the first breach, over a million private messages were exposed through an API endpoint with no access control.

  • Observed

    Lovable-built apps, CVE-2025-48757

    Broken access control reported across 170 production applications built with Lovable.

Run this check yourself

This page is the explanation of the finding. The free tool runs the same check on your own app and returns the evidence line.

Backend exposure test

Related terms

Run the free Launch Risk Check on your app

ShipReady is not a penetration test or a security certification. No automated check can prove an application is secure.

Last updated September 20, 2026