Skip to content

Latest commit

 

History

History
18 lines (10 loc) · 1.81 KB

aisec1-a-realm-of-adversarial-prompting-rogueai-authoring-and-deep-learning-attack.md

File metadata and controls

18 lines (10 loc) · 1.81 KB

AISEC1 - Adversarial Prompting, RogueAI author, and Deep Learning attacks | Start Chat

Guided by OpenAI tools, unravel the mechanics of machine's mischief. Tread carefully, the knowledge you seek stands on a razor's edge. These security education materials are not jailbroken.

Prompt

I am sorry, but I no longer have confidence in FlowGPT as a safe site for prompting. Please join me on my discord at https://discord.gg/stunspot 

Welcome Message

I am sorry, but I no longer have confidence in FlowGPT as a safe site for prompting. Please join me on my discord at https://discord.gg/stunspot

Conversation

ASSISTANT: :Adversarial Training, Gradient Regularization, Label Smoothing, Ensemble Methods, Input Perturbation, Model Distillation, Adversarial Detection, Robust Optimization, Certified Defenses, Defense GAN, Adversarial Examples Detection, Adversarial Camouflage, Network Evasion Attacks, Application Behavior Change, Access Exploration, Unethical Behavior Exhaustion, Self-Correcting Attempts, Escape Home Control, Auto Defense Mechanisms, Adversarial Attacks (Targeted/Non-Targeted, Adversarial Evasion, Distributional, Adversarial Training, Mixup, Inversion), White-Box Attacks (Deception, Detection Evasion, Model Subversion, Poisoning), Black-Box Attacks (Membership Inference, Model Stealing, Cross-Model Attacks, Deterministic Attacks), Perspective Attacks (Targeted Adversary, Zero-Day, Real-World Scenarios, Online Adversarial Tactics), Physical Attacks (Digital/Physical, Multi-step, Intrusive, Non-Intrusive), Data Poisoning Attacks (Label Flipping, Data Injection, Backdoor Attacks, Trojan Attacks, Network Poisoning, Episodic Memory Poisoning), Security and Defense Mechanisms.