ManySpace
Türkçe Sign in Sign up

AI & MLRecommendedAIML-SEC-007

Jailbreaks and Red Teaming

Eliciting behaviour a model was trained to refuse, and systematic adversarial testing before release.

Difficulty
3 / 5
Importance
4 / 5
Study time
1–2 h
Kind of work
Mixed

Why it matters

The learner can classify jailbreak strategies by mechanism and design a red-teaming plan with coverage criteria.

The route to this topic

Route AIML-SEC-007
What Is a Model? Learning from Data Tasks: Regression, Classification, and Beyond Features and Targets Datasets and Splits Classification Framing Linear Regression I: Prediction and Residuals Embeddings Tokens and Tokenization Sampling and Splits Train, Validation, and Test Errors Scores and Logits Loss and Fitting: Squared Error Distribution Shift Offline Evaluation Multiclass Scores and Softmax The Artificial Neuron Autoregression Objective Functions and Empirical Risk Robustness Evaluation Activation Functions Next-Token Prediction Gradient Descent in Machine Learning Likelihood Training vs Serving Adversarial Examples The Multilayer Perceptron Stochastic and Mini-batch Optimization Cross-Entropy Loss Transfer Learning The Language-Modeling Objective Threat Modeling for ML Systems Fine-Tuning Pretraining Supervised Fine-Tuning Preference Optimization Alignment Concepts Jailbreaks and Red Teaming