I. APPLICATION

This AUP applies to:

• Every authenticated user of runinfra.ai.
• Every pipeline created or deployed through the Service.
• Every model selected from the model registry and run on our managed GPU infrastructure.
• Every API request to your deployed endpoints, the unified inference proxy, and the optimization engine.
• Every output generated by a model running on the platform, whether displayed in the dashboard or returned via API.

You are responsible for the conduct of anyone who uses your account or your deployed endpoints, including end users of an API you built on top of the Service.

II. PROHIBITED CONTENT

You may not use the Service to generate, store, request, or distribute content that:

• Is unlawful under any jurisdiction that applies to you, us, or your end users.
• Depicts or sexualizes minors in any form. Child sexual abuse material is reported to NCMEC and the appropriate authorities; accounts are terminated immediately and may be reported to law enforcement.
• Threatens, harasses, or incites violence against a specific person or group.
• Promotes terrorism, violent extremism, or organized criminal activity.
• Generates non-consensual intimate imagery or deepfakes of identifiable real people without their consent.
• Facilitates fraud, phishing, identity theft, account takeover, scam pages, or counterfeit goods.
• Distributes malware, ransomware, exploits, or instructions for compromising systems you do not own.
• Infringes copyright, trademark, trade secret, patent, or publicity rights of any third party.
• Generates content presented as authored by a specific real person or entity in order to defame, defraud, or politically misrepresent them.
• Provides specific operational guidance for creating biological, chemical, radiological, or nuclear weapons capable of mass casualties.

III. PROHIBITED ACTIVITIES

You may not:

• Attempt to circumvent rate limits, billing meters, plan-tier restrictions, or compute quotas, including by spinning up multiple accounts to bypass per-account limits.
• Resell, sublicense, or proxy access to managed GPUs without a written reseller agreement with us.
• Train new models or fine-tune third-party models using the platform's compute for the purpose of building a competing inference-optimization product.
• Use the Service to scan, probe, or load-test infrastructure you do not own or have explicit permission to test, including any RunInfra infrastructure.
• Reverse-engineer, decompile, or extract the weights, kernels, or runtime artifacts of any model loaded onto the platform other than those you uploaded yourself.
• Misrepresent the Service's capabilities or outputs to third parties, including by claiming that we endorse, certify, or guarantee any model's behavior.
• Interfere with another user's Service, including by intentionally exhausting shared resources, abusing the inference proxy, or making automated submissions that disrupt scheduling.
• Use the Service in any way that violates the laws of the United States or the jurisdiction where you operate.

IV. AI-SPECIFIC RESTRICTIONS

The Service deploys third-party open-source models and routes inference through partner providers. The following AI-specific rules apply on top of the general restrictions above:

• Model license inheritance. You are responsible for complying with the license of every model you deploy. Many open-weight models carry use restrictions (no military use, no high-risk medical decisions, attribution requirements, etc.). We do not waive those licenses on your behalf.
• No covert AI. If your deployed pipeline generates content that will be shown to a human end user, you must not represent the output as written by a human when asked directly. You are not required to label every response, but you must not affirmatively deny that AI is involved.
• No automated critical decisions without a human. You may not use the Service's outputs as the sole basis for decisions that materially affect a person's legal status, employment, housing, credit, insurance, education, or health care without a meaningful human review step.
• No election manipulation. No generating fake election content, impersonating candidates or officials, or producing synthetic media of real political figures saying or doing things they did not say or do.
• Output ownership and responsibility. You own the outputs your deployed pipeline generates, subject to the model's underlying license. You are responsible for what those outputs do once they leave the platform.

V. FAIR USE OF COMPUTE

Each plan tier carries explicit limits on profiling sessions, optimization runs, deployed endpoint hours, and request volume. Those limits are published on the pricing page and enforced by the billing system. Beyond the numeric limits, “fair use” means:

• Optimization runs must serve a real intent to ship a pipeline. Repeatedly cancelling and restarting jobs to keep GPUs warm without making forward progress is not fair use.
• Deployed endpoints must serve legitimate inference traffic. Idle endpoints holding GPU capacity hostage may be cold-started by us to recover the resource.
• The unified inference proxy is for production traffic. Looping the proxy to scrape model behavior, probe other providers' rate limits, or build evaluation harnesses against our partners is not fair use.

When usage is abusive we may rate-limit or suspend the account before any monetary cap is hit. Where possible we contact you first.

VI. REPORTING VIOLATIONS

To report content or activity that violates this AUP, email legal@runinfra.ai. Include enough detail for us to find the offending pipeline, deployment, or output: the URL or pipeline ID if you have it, the date, and a description of the violation. We treat reports confidentially and do not share the reporter's identity with the reported user unless legally compelled.

For copyright takedown requests, follow the DMCA process described in our Terms of Service.

For security vulnerabilities, email security@runinfra.ai instead. Do not file vulnerabilities through public channels.

VII. ENFORCEMENT

When we believe in good faith that this AUP has been violated we may, at our discretion and in any order:

• Remove or quarantine the offending content, pipeline, or deployed endpoint.
• Rate-limit or pause your account.
• Require you to certify compliance before service is restored.
• Suspend the account pending investigation.
• Terminate the account.
• Report the activity to law enforcement when the conduct is criminal or there is a credible risk of imminent harm.

We try to give notice before taking enforcement action when the violation does not present an immediate risk. For high-severity violations (CSAM, credible threats of violence, active intrusion attempts) we act immediately and notify you afterward.

VIII. UPDATES

We may update this AUP to reflect new product capabilities, new legal requirements, or new abuse patterns. Material changes will be announced via the platform or by email at least 14 days before they take effect. Continued use of the Service after a change takes effect constitutes acceptance of the updated AUP.

IX. CONTACT US

Questions about this AUP, including borderline use cases you want to clear before launching, can be sent to legal@runinfra.ai.

RightNow AI, Inc.
131 Continental Dr
Newark, DE 19713
United States