Does anyone else worry about online exposure of AI-generated materials?

I am setting up a small PsychoPy Builder study for a lab project here in North Dakota, mostly simple routines and a few custom code components. I keep materials and screenshots in shared folders so my collaborators can check the flow before we run participants.

After reading about reports of many AI-made apps being exposed online, I started wondering about my own habits. Some of the files I use while drafting routines were produced with help from an AI assistant, and I am not sure how careful I need to be about where those drafts live or what gets uploaded. I do not want to overreact, but I also do not want participant-facing content sitting somewhere unintended.

For those of you building experiments in Builder, how do you decide what stays local versus in shared or cloud storage? Do you have a simple routine for checking exposure before a study goes live? I would appreciate hearing what has worked for others.

[Link to unknown news site removed by Wake. Read the original blog post by security firm Claroc about apps hosted by Grok here: Grok Apps Exposed xAI Secrets and Private Data | Claroc Blog]

As I understand it, users uploaded personal data to their app space, presumably thinking it was secure.

Personally, I am not concerned about uploading code to an AI, though I do ensure I use one approved by my employer. I would not upload participant data, or personal information.

Where Pavlovia users should be concerned is that data from projects set to internal can be viewed by any logged in user (public is no longer available as an option). Database data is only visible to users with explicit permission so that is my recommended route for collecting data for projects which you might want to share with others.

Participant and personal data in private cloud storage accounts should be secure. I don’t have concerns with data saved on Google Drive or Dropbox, whereas I try to minimise the amount of data kept in mysql databases.