AI & Machine Learning

Understand the machines rewriting everything

Novice

Mastery0%

Novice

Note. AI tools change fast and confidently get things wrong. Verify anything high-stakes against primary sources, and never paste secrets/PII into untrusted endpoints.

Concept 17 of 19

debated

Prompt injection

User content can hijack the model just like SQL injection hijacks a database.

Definition

A class of attacks where adversarial instructions embedded in untrusted input cause a language model to disregard its developer's intended instructions.

Step 1

Predict first

In one or two sentences, what do you think this concept means? Wrong is fine; the small struggle is what makes the answer stick.