Acceptable Use Policy

01Overview

This Acceptable Use Policy ("AUP") applies to everyone who uses AirLLM's Services — as defined in our Terms & Conditions, including Air Router, AirGPU, AirCode CLI, AirCode Studio, the Control Plane, and Alina — and to any product you build on them. It exists to keep the platform safe for our customers, the upstream model providers we route to, and the public. It is incorporated into our Terms & Conditions.

Two rules sit above everything else: comply with the law, and don't use the Services to harm people. Because completions are generated by third-party models, the usage policies of the provider serving your request also apply to that request. Anything an AirCode agent does at your direction is treated as your own use of the Services.

02Universal prohibitions

You may never use the Services to:

Endanger children

• Create, request, or distribute child sexual abuse material, sexualized depictions of minors, or any content that grooms, exploits, or endangers a minor. We report apparent violations to the National Center for Missing & Exploited Children and cooperate with law enforcement.

Facilitate violence or weapons

• Develop, produce, or procure weapons — including chemical, biological, radiological, nuclear, or high-yield explosive capabilities — or provide uplift toward doing so;
• Plan, incite, threaten, or glorify violence, terrorism, or human trafficking.

Conduct malicious cyber activity

• Create or deploy malware, ransomware, or exploits; conduct unauthorized access, denial-of-service attacks, or credential theft; or build phishing and social-engineering infrastructure. Good-faith security research on systems you are authorized to test is permitted.

Violate privacy or enable surveillance

• Track, monitor, or profile individuals without lawful basis; build facial-recognition databases from scraped imagery; facilitate stalking or doxxing; or process biometric or other sensitive data unlawfully.

Deceive or defraud

• Run scams, generate fraudulent documents or reviews, impersonate real people or organizations, produce coordinated disinformation, or interfere with elections and civic processes — including misleading content about how, when, or where to vote.

Abuse or harass

• Generate content that harasses, threatens, demeans, or incites hatred against people, including on the basis of protected characteristics; or create non-consensual intimate imagery of real people.

Break the law generally

• Facilitate any other illegal activity, including trafficking in illegal goods, money laundering, sanctions evasion, or infringement of others' intellectual-property rights. To report copyright infringement, use the DMCA process in Section 26 of our Terms & Conditions.

03High-risk uses & required safeguards

Some legitimate uses carry elevated risk to people's health, rights, or finances. You may pursue them on AirLLM only with the following safeguards in place:

04Platform integrity

To keep the gateway fair and reliable for everyone, you must not:

• Circumvent rate limits, usage budgets, loop protection, safety filters, or the fail-closed billing checks;
• Share, sell, or pool API keys across organizations, or resell raw access to the Services without a reseller agreement;
• Probe, scan, or stress-test the Services except through our authorized security-testing process (request authorization through our Contact page);
• Scrape the Services, harvest other customers' data, or attempt to extract non-public routing logic or model configurations;
• Use the Services or Output to develop or train AI models that compete with AirLLM's Services, or in any way the serving provider's terms prohibit;
• Misrepresent benchmark results by publishing comparisons without disclosing the alias and configuration used.

05Responsible use of AirCode agents

AirCode CLI and AirCode Studio can read, create, change, and delete files, run commands, install dependencies, make commits, launch simulators, emulators, and browsers, and reach network resources on your machine at your direction. When you run them:

• You supervise the agent. Review its changes and commands, and stay accountable for everything it does under your account, as described in the Terms;
• Point agents only at code, systems, accounts, and networks you own or are authorized to use, and never use them to access, scan, or change third-party systems without permission;
• Do not direct an agent to do anything this policy prohibits — an action an agent takes at your direction is treated as your action;
• Do not use agents to evade rate limits, usage budgets, loop protection, or other safeguards, or to generate excessive automated load;
• Keep your work under version control and backed up, and run agents in environments suited to the task, away from production systems and credentials you cannot afford to have changed or exposed.

06AI disclosure requirements

If you build user-facing products on AirLLM:

• Disclose that users are interacting with an AI system wherever a reasonable person could otherwise believe they are talking to a human — especially in chat, voice, support, and companionship contexts;
• Do not present generated content as human-authored where authorship materially matters (journalism, reviews, academic work, legal filings);
• Comply with AI-transparency laws applicable to your users, including the EU AI Act's transparency obligations where they apply to your deployment.

07Reporting & enforcement

Report suspected violations through our Contact page — including if you believe your own systems or keys were compromised and misused. Copyright complaints go to our Copyright Agent, as Section 26 of our Terms & Conditions describes.

We enforce this AUP with automated abuse detection and human review. Enforcement is at AirLLM's discretion: we generally act in proportion to the violation, but we may take any of these steps, in any order, including immediately:

1. Warning — for minor or apparently inadvertent violations, with guidance on how to remediate;
2. Throttling or feature limits — reduced rate limits or restricted model access while an issue is investigated;
3. Suspension — immediate suspension for serious violations or ongoing harm;
4. Termination — account termination for egregious or repeat violations, per the Terms.

Where we take enforcement action, we will tell you what happened and how to appeal — appeals go to our Legal Cell at legal@airllm.ai or through our Contact page — unless doing so would compromise safety, security, or a legal obligation. If a safety guardrail or an Acceptable Use review is triggered, we may keep a record that includes the text that triggered it, for up to 2 years, for enforcement and legal defense. Information associated with confirmed violations — such as account and usage records and any content reported to us — may also be retained under the safety-hold provisions of our Data Retention Policy, and we may notify upstream providers or law enforcement where legally required or where there is a risk of serious harm.

We are not obligated to monitor use of the Services. To the maximum extent permitted by applicable law, the AirLLM Parties (as defined in the Terms) are not liable to you or anyone else for any enforcement action we take, or decline to take, under this AUP, and suspension or termination for a violation does not entitle you to a refund.