Good question, and fair point, the title can make it sound more active than it is.
What I meant was, AI doesn’t go out and read sensitive data on its own. It processes whatever it’s given. The issue is that in real workflows, people often paste or connect sensitive data into these systems like customer info, internal docs, financial data, etc.
Once that happens, the data can flow through prompts, logs, APIs, or outputs. That’s where the risk comes from, not the model seeking data, but how easily sensitive data gets pulled into the system and propagated.
Very useful information thank you suny!
I’m glad I could help!
Where did you check that AI reads sensitive data ?
Good question, and fair point, the title can make it sound more active than it is.
What I meant was, AI doesn’t go out and read sensitive data on its own. It processes whatever it’s given. The issue is that in real workflows, people often paste or connect sensitive data into these systems like customer info, internal docs, financial data, etc.
Once that happens, the data can flow through prompts, logs, APIs, or outputs. That’s where the risk comes from, not the model seeking data, but how easily sensitive data gets pulled into the system and propagated.