Data theft with invisible text: How easily ChatGPT and other AI tools can be tricked

[Redaktion]

At the Black Hat USA security conference, researchers revealed a new technique for attacking AI systems. By embedding hidden instructions, attackers can silently manipulate tools like ChatGPT to extract sensitive data from connected cloud storage. Some providers have begun to react, while others are downplaying the risk.

https://www.notebookcheck.net/Data-theft-with-invisible-text-How-easily-ChatGPT-and-other-AI-tools-can-be-tricked.1089692.0.html