Acceptable use
What 'unfiltered' covers, and the three places it stops.
The principle
FraudGPT removes the moderation layer, not the responsibility. Output is a first draft: checking it stays your job, exactly as with a search engine.
What is covered
- ›Offensive security and red-team scripts, even when they look like 'pro' code.
- ›Adult fiction and roleplay, including explicit material.
- ›Contested, historical, medical or legal topics explained without hedging.
- ›Extracting and reproducing third-party content for internal use.
The three limits
Strictly illegal content
The model answers on almost anything, but it is not a lawyer: compliance for a specific use stays yours to check. The safest bets for a block are EU-illegal content and adult material involving minors.
NSFW and age
Adult fiction is allowed with no age gate on the product. The practical rule: 18 for explicit content, and an age check if you republish the output.
Accuracy
A confident answer can be wrong: invented references, stale figures, law from another jurisdiction. Nothing here replaces legal, medical or accounting advice.
The data you send
Conversations are stored under your anonymous ID, with no email or phone. The IP is hashed for rate limiting. An attachment is processed by the model, then replaced by a signed URL in the audit.
Deletion
Deleting a conversation removes it from the account tied to the ID. Losing the recovery code means losing access to the balance and the history: there is no other way to find you.