Acceptable Use Policy.

Effective date: July 1, 2026 · Keld Inc

This Acceptable Use Policy ("AUP") sets the rules for using Keld. It's part of our Terms of Service and applies to everyone who uses the marketplace, APIs, SDKs and related products. Providers on the supply side must also follow it. Breaking it can lead to suspension or termination.

Prohibited content and activities

You may not use Keld to create, process, distribute or enable any of the following:

• Illegal content or activity — anything unlawful where you or your users are, including content that violates sanctions or export controls.
• Child sexual abuse material (CSAM) or any sexual content involving minors — zero tolerance; we report as required by law.
• Exploitation and serious harm — content that promotes terrorism or violent extremism, facilitates human trafficking, or incites violence.
• Dangerous capabilities — instructions to build weapons or to develop chemical, biological, radiological or nuclear harms; serious facilitation of physical harm.
• Malware and intrusion — creating or distributing malicious code, or attempting to breach, disrupt, overload, or probe the security of Keld, providers or third parties.
• Fraud and deception — phishing, scams, impersonation, fake identities, or coordinated inauthentic behavior.
• IP and confidentiality violations — infringing copyrights, trademarks, patents or trade secrets, or misusing others' confidential information.
• Privacy violations — unlawful processing of personal data, doxxing, or unauthorized surveillance.
• Hate and harassment — content that harasses, threatens, or demeans people based on protected characteristics.
• Spam and abuse of resources — bulk unsolicited messaging, scraping in breach of terms, or evading rate limits, quotas, or usage controls.
• High-risk automated decisions — using model output as the sole basis for decisions with legal or safety consequences (e.g. medical, legal, financial, employment, or critical-infrastructure decisions) without appropriate human review and safeguards.

Individual model providers may set additional usage policies. When your job runs on a provider, you must also comply with that provider's published use rules.

You are responsible for the data you submit

You represent that, for everything you send through Keld for inference, you have the necessary rights, permissions and lawful basis — and that the data and your intended use comply with applicable law and this AUP. You're also responsible for how you use the output you receive.

Encrypted payloads and our role

Keld is built to protect the confidentiality of your jobs. The contents of inference payloads are encrypted in transit and run under Zero Data Retention, and Keld does not inspect, monitor or filter the content of those payloads. That design protects you — but it does not move responsibility for the content from you to Keld. You remain responsible for ensuring your use complies with this AUP and the law, whether or not Keld can see the content.

Where we receive a credible report or valid legal process, or where we detect abuse from metadata or platform signals, we may investigate and act — including suspending access, removing access to a feature, or reporting unlawful activity to the appropriate authorities.

Enforcement

We may suspend or terminate access, with or without notice depending on severity, for violations of this AUP or risks to the Service, providers, or others. We'll use reasonable judgment and, where practical, give you a chance to fix the issue.

Reporting abuse

To report a suspected violation, email compliance@keld.co with details.