SESKStuff Everyone Should Know
SESK: Stuff Everyone Should Know · sesk.info/ai-jailbreak-system-prompt

What does an AI jailbreak try to get around?

The model's restrictions on its outputs, such as refusals.

Start playing free

  1. Restrictions on what it will output ✔
  2. Its training dataset
  3. Its tokenizer vocabulary

Why: NIST defines a jailbreak as a direct prompting attack meant to get around limits on a model's outputs, such as its refusals. Attackers can also use one to override the system prompt.

Huh, didn't know that: Benchmarks such as JailbreakBench and HarmBench measure how easily models can be jailbroken.

The dad joke
The jailbreak brought a fake hall pass to the system prompt.
Like this card?
Play the game

References

Answer: Restrictions on what it will output. NIST defines a jailbreak as a direct prompting attack meant to get around limits on a model's outputs, such as its refusals. Attackers can also use one to override the system prompt.

  1. NIST Adversarial Machine Learning Taxonomy nvlpubs.nist.gov/nistpubs/ai/NIST.AI.100-2e2025.pdf
    JAILBREAK is a direct prompting attack intended to circumvent restrictions placed on model outputs
    Checked 2026-10-10.

Think this answer is wrong? Tap "Challenge this answer" on the card in the game and tell us why.

Test yourself in the game

Artificial IntelligenceWhat is a common use of the system prompt in a chat model API?Artificial IntelligenceHow do AI, machine learning and deep learning fit together?Artificial IntelligenceWhat did Alan Turing call his 1950 test of machine intelligence?

That's stuff everyone should know.