Last updated: April 6, 2026

Responsible Use Policy

This Responsible Use Policy sets expectations for how the inference-relay library and related services (the “Service”) may and may not be used. It supplements our Terms & Conditions.

1. Purpose

inference-relay is routing infrastructure. It connects your code to your AI providers. It does not generate, filter, moderate, or inspect the content that passes through it. Because we are a routing layer, the responsibility for what is routed rests with you. This policy defines the boundaries of acceptable use to protect the integrity of the Service, the safety of the broader ecosystem, and compliance with applicable law.

2. Permitted Uses

The Service is intended for the following lawful applications:

  • Routing AI inference requests for lawful software applications and services.
  • Building development tools, internal automation, and productivity workflows.
  • Research, analysis, and content generation for lawful purposes.
  • Educational and academic applications.
  • Personal and organizational automation using your own AI subscriptions.

3. Prohibited Uses

You may not use the Service for, or in connection with, any of the following:

Illegal Activity
Any activity that violates applicable local, national, or international law or regulation.
Circumventing AI Provider Terms
Using the Service to circumvent, bypass, or violate the terms of service, acceptable use policies, or usage limits of any AI provider, including but not limited to Anthropic and OpenAI.
Harmful Content Generation
Routing requests designed to generate child sexual abuse material (CSAM), weapons manufacturing instructions, credible threats of violence, or targeted harassment campaigns.
Unauthorized Access
Accessing, using, or attempting to use another person's or entity's AI provider credentials, API keys, or subscriptions without their express authorization.
Credential Sharing and Resale
Sharing, reselling, sublicensing, or otherwise distributing your inference-relay license key, or using a single license key beyond the seat limits of your subscription tier.
Automated Abuse
Automated scraping of the Service or its infrastructure, deliberate circumvention of rate limits, or any activity designed to degrade the Service's availability or performance.
Competing Product Development
Using the Service to develop, benchmark, train, or improve any product or service that competes with inference-relay, including AI inference routing, relay, or orchestration libraries. This includes using the Service to reverse engineer inference-relay's routing algorithms, fallback strategies, or instruction protocol.
Misrepresentation
Misrepresenting the capabilities, origin, or nature of the Service to third parties. You may not claim that inference-relay provides AI generation, content moderation, or safety filtering services.

4. Developer Responsibilities

If you integrate inference-relay into an application or service that is used by end users, you accept the following responsibilities:

  • Transparency. Inform your end users that their AI inference requests are routed through the relay infrastructure. Disclose the nature of the routing to the extent required by applicable law.
  • Consent. Obtain any consent required by applicable law before routing end user requests through the Service, particularly where the end user’s own AI subscription is used for execution.
  • Provider compliance. Ensure that your application’s use of the Service complies with the acceptable use policies and terms of service of every AI provider you route through, including Anthropic’s Acceptable Use Policy and OpenAI’s Usage Policies.
  • Content responsibility. You are solely responsible for all content routed through the Service using your account. Dumbell2 does not monitor, filter, or moderate content routed through inference-relay.
  • Telemetry integrity. Do not tamper with, forge, or suppress telemetry data. Do not manipulate token counts or cost calculations.

5. Enforcement

Violations of this policy are handled through a graduated enforcement process:

1
Warning
For first-time or minor violations, we will notify you of the violation and request corrective action within a specified timeframe.
2
Suspension
For repeated or serious violations, we may suspend your access to the Service while we investigate. During suspension, your license key is temporarily deactivated.
3
Termination
For egregious or continuing violations, we will terminate your account and permanently revoke your license. No refund is provided for termination due to violation.
4
Legal Action
In cases involving illegal activity, intellectual property theft, or material harm, Dumbell2 reserves the right to pursue legal remedies, including cooperation with law enforcement authorities.

Dumbell2 reserves the right to skip steps in this enforcement ladder when, in its sole discretion, the severity of the violation warrants immediate action.

6. Reporting Violations

If you become aware of a violation of this policy, please report it to:

Include as much detail as possible, including the nature of the violation, any relevant account identifiers, and supporting evidence. We review all reports and will respond within 5 business days.