Ok but it’s not like pg can’t stop you from doing something dumb.

There are probably countless new projects today that are storing plaintext passwords, or not adding scoping, and so on.

Putting in scopes and ensuring data security for both users and system wide is on the developer.

It's hard to screw up Postgres to the extent that your entire DB is made fully accessible by all users. This has happened many times with Firebase apps, for over a decade.

You could have a SQL injection vuln, but any SQL lib will very clearly steer you to parameterized queries, and even then such a vuln takes some expertise to find and exploit.

That’s simply not true. I remember working with a startup founder who had Jerry rigged some crap shit together with gpt a year ago.

I was able to access his data by simply accessing it figuring out his URL and other stuff. I told him to use supabase or DO deployment and set up proper roles and stuff…

I think you’re being way too charitable honestly and it’s dangerous. I won’t join you on that path of absolving the developer of any blame.

They don’t read the docs and they didn’t care simply put. Any production system needs to be tested especially if it will have PII data.

Don't get me wrong, there is still such a thing as a bad dev. If this startup founder actually wrote an entire app using GPT and it had such vulns, I'm pretty sure he'd mess up the Firebase ACLs too.