We haven't found a way to prevent AI chat systems from being manipulated into revealing internal company data through clever prompting.
open
Global / Unspecified, Global
Employees or attackers can sometimes use carefully crafted prompts to trick AI systems integrated into company tools into revealing sensitive internal information they weren't meant to share. Fully closing this security gap remains an unresolved AI safety challenge.
Citation ID: WS00819
Title: We haven't found a way to prevent AI chat systems from being manipulated into revealing internal company data through clever prompting.
URL: https://worldsolve.org/index.php?api=problem&id=819