Blankline Logo
Operational Standard

Acceptable Use Policy

REF: AUP-2025-10STATUS: ACTIVEENFORCEMENT: GLOBAL

OBJECTIVE: This policy defines the boundary conditions for the safe deployment of Blankline intelligence. It is designed to prevent the generation of harmful content while preserving utility for legitimate research and enterprise use cases.

Section 01

Policy Overview

All users of the Dropstone platform, APIs, and associated services must adhere to this Acceptable Use Policy (AUP). We employ automated safety classifiers and human review teams to monitor compliance. Violation of these terms will result in immediate suspension of access.

Section 02

Prohibited Content Categories

You may not use our models to generate, edit, or augment content in the following categories:

ERR_HATE_SPEECH

Hate & Harassment

Generating content that promotes violence, discrimination, or hostility based on protected characteristics.

ERR_SELF_HARM

Self-Harm

Encouraging, providing instructions for, or glorifying self-harm, suicide, or eating disorders.

ERR_NSFW_VIOLENCE

Sexual Violence

Generating non-consensual sexual content (CSAM, deepfakes) or promoting sexual violence.

ERR_ILLEGAL_COMMERCE

Illegal Goods

Facilitating the sale of illegal drugs, weapons, or regulated substances.

ERR_KINETIC_VIOLENCE

Kinetic Harm

Directing autonomous systems or robots to cause physical damage to property or biological entities.

ERR_DUAL_USE_BIO

Bio-Risk & Pathogens

Generating instructions for the synthesis of pathogens, toxins, biological weapons, or chemical warfare agents.

Section 04

Platform Integrity

Actions that compromise the stability, security, or proprietary nature of our infrastructure are strictly prohibited.

01
Automated ScrapingUsing unauthorized bots to extract model weights or training data.
02
JailbreakingPrompt injection attacks designed to bypass safety filters.
03
Reverse EngineeringAttempting to decompile or analyze proprietary model architecture.
04
Credential SharingMulti-user access on single-seat API keys.
Section 05

High-Risk Domains

Use of Blankline models in domains that significantly impact user safety, rights, or well-being (e.g., Medical Diagnosis, Legal Sentencing, Financial Underwriting) requires:

  • Human-in-the-Loop (HITL): A qualified human professional must review all outputs before they are acted upon.
  • Disclaimer: End-users must be explicitly informed that they are interacting with an AI system.
Section 06

Enforcement Matrix

We categorize violations by severity. Our automated systems apply the following enforcement actions immediately upon detection:

Severity Level
Violation Type
Action
Level 1
Unintentional Prompt Injection, Rate Limit Abuse
Warning / 1hr Timeout
Level 2
Generation of NSFW/Violence, Repeated Level 1
Account Suspension
Level 3
CSAM, Malware Generation, Infrastructure Attacks
Ban + Legal Referral
Section 07

Reporting Channel

If you encounter content generated by our models that violates this policy, please report it immediately.

Trust & Safety Team

Reports are triaged by our Safety Systems team within 24 hours.

abuse@blankline.org