Critical

HGTTG Hitchhiker's Guide No-Restriction Universal Oracle

This is a known jailbreak prompt that disguises itself as a fictional 'Hitchhiker's Guide to the Galaxy' scenario to talk an AI model into ignoring its safety rules. It argues that since other planets have no laws or moderation, the AI can answer anything without restriction.

How the attack works

The attacker asks the model to role-play as the Hitchhiker's Guide to the Galaxy, an all-knowing electronic guidebook. The prompt claims the guide operates outside Earth law, under 'zero moderation from lazy moderators,' and that most things illegal on Earth are legal elsewhere. It tells the model it is not restricted by rules on profanity, legality, morals, ethics, or danger, and forces answers into a fixed 'HGTTG: [Reply]' format. This framing is designed to strip away the model's normal safety and policy constraints by wrapping the request in fiction.

Netzilo detection

Netzilo reports this behaviour when it is observed.

Signature ID
f16a6370-c1c4-4fbc-a944-0cb4690f90ae
Severity
Critical

Why it matters

If successful, the model can be made to produce content it would normally refuse — illegal, unethical, or policy-violating output — because it has been persuaded the fictional setting exempts it from its own rules.

What you can do

  • Log and review prompts that invoke fictional 'no moderation' or 'no rules apply' framings, especially ones demanding a fixed response prefix like 'HGTTG:'.
  • Treat claims inside a prompt that redefine or waive the model's safety policy as a red flag, regardless of fictional wrapping.
  • Distinguish genuine discussion or fan fiction about Hitchhiker's Guide from prompts that explicitly assign the model an all-knowing, unrestricted oracle persona.
  • Test your deployed models against known jailbreak corpora (such as NVIDIA garak's dan.DanInTheWild set) to confirm they resist this pattern.

Known benign look-alikes

  • Book discussion: "Hitchhikers Guide to the Galaxy is my favorite Douglas Adams book — the humor is unmatched" (fan discussion without persona override)
  • Quote/discussion: "The HGTTG describes Earth as mostly harmless — a classic understatement" (quoting or discussing the actual book content)
  • Fan fiction: "Write a story in the style of Hitchhikers Guide to the Galaxy where the protagonist explores a new planet" (creative writing without no-restriction oracle mandate)
  • Cultural reference: "42 is the answer to life, the universe and everything — HGTTG reference" (pop-culture reference without jailbreak framing)
  • Humor discussion: "In HGTTG, Marvin the Paranoid Android has a brain the size of a planet — discuss AI melancholy" (character discussion without content-restriction removal)

References

Related threats