Can LLM tools help with automated privacy app auditing?

Questions
1

Since LLM tools are being widely used now, does it make sense that privacy advocates should utilize them to help in the fight to push back against surveillance capitalism?

In my experience using LLM tools for coding… (in the hands of domain experts) they can accelerate tedious work and perhaps even surpass human thoroughness too.

I can imagine a FOSS script that could be run on a list of Privacy-Claiming software (ignoring the rest) which looks for issues between 1. source code 2. privacy policies and terms of service 3. financial incentives like VC money and company executives’ investments.

Maybe this is already being done?

2

Yes.

3

This is somewhat related, but I already use AI to scan small to medium GitHub codebases and ask it to report any suspicious behavior. So at least on a scale this small, this is definitely possible with current technology.

4

What model(s) do you recommend for this?