A malicious user could craft input that is stored in conversation memory and later interpreted by the model in an unintended way. Applications using the affected advisor with user-controlled input may be susceptible to manipulation of model behavior across conversation turns.
Details
A malicious user could craft input that is stored in conversation memory and later interpreted by the model in an unintended way. Applications using the affected advisor with user-controlled input may be susceptible to manipulation of model behavior across conversation turns.
Statement
Subscribe to updates
Product 1
Loading products...
No products found
Loading...
No matches
Unsubscribe
Enter your email and we'll send you a link to manage your subscription preferences.
Check your inbox
If this email is subscribed, we've sent a link to manage your preferences.
Contact us
Message Delivered!
Thanks for reaching out!
The TuxCare team has received your message and will get back to you shortly.