Topic: contrast

Contrast Security adds new feature to help protect against prompt injection in LLMs

Prompt injection — attacks that involve inserting something malicious into an LLM prompt to get an application to execute unauthorized code — topped the recently released OWASP Top 10 for LLMs.  According to Contrast, this could result in an LLM outputting incorrect or malicious responses, producing malicious code, circumventing content filters, or leaking sensitive data. … continue reading

DMCA.com Protection Status