Prompt Laundering

ELI5 — The Vibe Check

Prompt laundering is rewording a prompt to get around an AI's refusal without actually changing what you're asking for. Mildly manipulative. Sometimes necessary (false refusals). Sometimes a jailbreak vector.

Real Talk

Prompt laundering is rephrasing a refused or filtered prompt to bypass model guardrails while retaining the original intent. Legitimate uses: working around false-positive refusals. Illegitimate uses: jailbreak attempts. Model providers increasingly detect laundering via intent classifiers that run before generation.

When You'll Hear This

"The model refused, so I prompt-laundered it into a hypothetical." / "Don't prompt-launder past safety — log it as a legit refusal."