Invisible text that AI chatbots understand and humans can’t? Yep, it’s a thing.

stapp · Oct 16, 2024

What if there was a way to sneak malicious instructions into Claude, Copilot, or other top-name AI chatbots and get confidential data out of them by using characters large language models can recognize and their human users can’t? As it turns out, there was—and in some cases still is.
The invisible characters, the result of a quirk in the Unicode text encoding standard, create an ideal covert channel that can make it easier for attackers to conceal malicious payloads fed into an LLM.
Click to expand...

Focusing more narrowly on AI security, the phenomenon of LLMs reading and writing invisible characters opens them to a range of possible attacks. It also complicates the advice LLM providers repeat over and over for end users to carefully double-check output for mistakes or the disclosure of sensitive information.
Click to expand...

https://arstechnica.com/security/20...isible-text-creating-an-ideal-covert-channel/

So not only do you have to read the things you can see, you have to be able to read the things you cannot. What a crazy world.

zapjb · Oct 16, 2024

It's official I'm now an AI Luddite.

Log in or Sign up

Invisible text that AI chatbots understand and humans can’t? Yep, it’s a thing.

stapp Global Moderator

zapjb Registered Member

Log in or Sign up

Invisible text that AI chatbots understand and humans can’t? Yep, it’s a thing.

stapp Global Moderator

zapjb Registered Member

Useful Searches