AdvancedVocabulary#ai-llm#security#developer-tools

AI Red Teaming Vocabulary

Build fluency in the vocabulary of proactively testing a model for a harmful or exploitable response.

0 / 5 completed
1 / 5
At standup, a dev mentions deliberately trying to manipulate a model into producing a harmful or policy-violating response before that model ever reaches real users. What is this practice called?