45
Slack AI can leak private data via prompt injection
(www.theregister.com)
c/cybersecurity is a community centered on the cybersecurity and information security profession. You can come here to discuss news, post something interesting, or just chat with others.
THE RULES
Instance Rules
Community Rules
If you ask someone to hack your "friends" socials you're just going to get banned so don't do that.
Learn about hacking
Other security-related communities !databreaches@lemmy.zip !netsec@lemmy.world !cybersecurity@lemmy.capebreton.social !securitynews@infosec.pub !netsec@links.hackliberty.org !cybersecurity@infosec.pub !pulse_of_truth@infosec.pub
Notable mention to !cybersecuritymemes@lemmy.world
Is it possible to implement a perfect guardrail on an AI model such that it will never ever spit out a certain piece of information? I feel like these models are so complex that you can always eventually find the perfect combination of words to circumvent any attempts to prevent prompt injection.
Reminded me of this game: https://gandalf.lakera.ai/intro