It's great to use Lovable to generate applications in one sentence, but whether the generated things are safe or not, this time it was carefully checked.

Researchers have revealed that several apps built using Lovable exposed users’ passport information and more than 200,000 private messages. Lovable's security director admitted during the discussion that the root cause was insufficient row-level security (RLS) configuration.

Explain what RLS is. For applications generated by Lovable, the backend is usually Supabase's Postgres database, and the frontend directly looks up tables through the Data API. Whether this architecture can protect data depends entirely on whether the RLS rules on each table are written and whether they are written correctly: without the rules, anyone who gets the front-end public key can read the entire table.

This type of problem is not unique to Lovable. All solutions that allow the front-end to directly connect to the database have the same pitfalls, except that in the past, the code was written by programmers, who knew to some extent that permissions were required; now the code is written by AI, and the people who use it may not know RLS at all.

Lovable's response is: the platform will automatically scan the application before it is released, and also provides all users with deep static code scanning (SAST) for free. If problems are found, they can be repaired with one click, or open the background for automatic repair.

However, there are no answers to several key questions: how many applications are affected, how long the data has been exposed, and whether old applications that have been released can be covered by the scan. If you have an application built with Lovable or similar tools that is already collecting user data, the safest way is to go to the Supabase backend now to check whether RLS is turned on table by table.