Quiz: How AI Attacks Actually Work

This is a member-only chapter. Log in with your Signal Over Noise membership email to continue.

Module 2 · Quiz

Check Your Understanding

Answer all questions correctly to complete this module.

Check Your Understanding

Answer all questions correctly to complete this module.

1. What is prompt injection?

Injecting malicious code into an AI model's source code Hiding malicious instructions inside content that an AI is asked to process Using AI to generate prompts that hack into databases A method of bypassing multi-factor authentication

2. What practical test does the chapter suggest can detect current real-time deepfake video?

Asking the person to wave their hand in front of their face Running commercial deepfake detection software Asking the person to slowly turn their head to a 45-degree angle and hold it Checking for inconsistencies in the audio quality

3. What is the common thread across all attack types described in this module?

They all require nation-state resources to execute They all exploit things people were trained to trust: familiar faces, voices, and professional writing They all target large corporations exclusively They all involve compromising AI models directly

Previous The Common Thread

Next Introduction