Acceptable Use Policy

Applies to all use of Orbital

This policy applies to Orbital, a product of Constructr, Inc. ("Orbital", "we", "us", "our").

Orbital runs AI coworkers that take real actions in real systems - your repositories, your inboxes, your CRM, your production services. That makes acceptable use a narrower question than it is for a chatbot. This policy sets out what you may not do with Orbital, what you may not have an agent do, and what happens when those lines are crossed.

This policy is part of the Terms of Service. If you break it, you break the Terms. Capitalised terms not defined here have the meaning given in the Terms of Service or in the AI & Agent Terms.

1. Who this applies to

This policy applies to you if you hold an Orbital account, to everyone you invite into your workspace, and to every agent you configure, run or connect. It applies to the orchestrator, to the agent harnesses you bring, and to anything an agent does through a connector.

You are responsible for your agents. An action an agent takes with the authority you gave it is your action. "The agent did it" is not a defence under this policy, and we will treat it the same way we would treat the same action taken by a person on your team.

2. Prohibited uses and content

You may not use Orbital, and may not permit anyone in your workspace or any agent to use Orbital, to:

2.1 Break the law or infringe rights

Do anything unlawful, or help anyone else do anything unlawful.
Infringe or misappropriate someone else's intellectual property, publicity rights or privacy rights.
Process personal information you do not have the right to process, or in a way that breaks the privacy notice you gave the people it belongs to.

2.2 Harm people

Generate, store or distribute material that sexually exploits or endangers children, including synthetic material.
Harass, threaten, defame, intimidate or stalk anyone.
Create or spread hate speech, or material that advocates violence, terrorism, violent extremism or human trafficking.
Encourage self-harm or suicide.
Produce sexually explicit material, or operate a sexual chat service.
Develop, market or distribute weapons, weapon accessories or explosives.

2.3 Use AI in ways we will not support

These prohibitions are deliberately drafted to a single standard worldwide, so that the policy does not need rewriting for each market we enter. You may not use Orbital to:

Deploy subliminal, manipulative or deceptive techniques intended or likely to distort someone's behaviour in a way that causes them significant harm.
Exploit the vulnerabilities of a person or group - because of age, disability, or a specific social or economic situation - in a way that causes or is likely to cause significant harm.
Score or classify people based on their social behaviour or inferred personality in a way that leads to detrimental treatment unrelated to the context in which the data was collected, or treatment disproportionate to the behaviour.
Predict or infer a person's protected characteristics - racial or ethnic origin, citizenship or immigration status, political opinions, religious or philosophical beliefs, union membership, sex life or sexual orientation, health status, disability, or criminal history or propensity. This does not stop you from using Orbital to detect fraud or security abuse, or to find and reduce bias in your own systems.
Build or expand facial-recognition databases by untargeted collection of facial images from the web or from CCTV footage.
Infer people's emotions in a workplace or an educational setting, except where it is strictly necessary for a medical or safety purpose.
Submit images, video, audio or biometric samples of people in order to create, analyse or categorise biometric identifiers.

2.4 Make significant decisions about people without a real human in the loop

You may not use Orbital to make, or to substantially replace a human in making, a decision that provides or denies someone: financial or lending services; housing; education enrolment, credentials or opportunities; employment or independent-contracting opportunities, work allocation, compensation, promotion, demotion, suspension or termination; or healthcare services.

You may use Orbital to support those decisions. If you do, the person who makes the decision must know how to interpret the agent's output, must review that output together with the other information relevant to the decision, and must have the authority to make or change the decision. A reviewer who approves whatever the agent proposes does not satisfy this requirement. Full detail, including the notice and appeal obligations this can trigger, is in AI & Agent Terms §6.

2.5 Give regulated or professional advice

You may not use Orbital to generate individualised advice that a licensed professional would normally give - legal, financial, tax, medical or psychological - for delivery to an end user.
You may use Orbital to assist a qualified professional, provided that professional reviews the output and remains accountable for it.
Do not use Orbital for medical diagnosis or treatment decisions, or for safety-critical control of physical systems.
Do not submit protected health information. Orbital is not configured for PHI and we do not offer a Business Associate Agreement.

2.6 Deceive people

You may not present an agent as a human being. If an agent interacts directly with a person outside your organisation, tell them they are dealing with an automated system.
You may not pass agent output off as human-written where that misrepresentation matters - in academic work, regulatory filings, legal submissions, or anywhere disclosure is required.
You may not impersonate a real person or organisation, including by voice or likeness, without their documented consent.
You may not use Orbital for fraud, scams, phishing, or fake reviews and engagement.

3. Prohibited agent behaviour

This section is about what an agent does, not what it writes. It applies whenever an agent acts through a connector, a browser, an API, a shell, or any other tool.

3.1 Only act in systems you control

You may point an agent at a system only if you own it or you have documented authority from its owner to act in it in the way the agent will act. "Documented" means you could produce the authorisation if asked.

You may not use Orbital as a general-purpose tool for taking automated action against systems belonging to anyone else - including competitors, suppliers, public bodies, or individuals.

3.2 Respect other people's terms, limits and controls

You may not configure or instruct an agent to:

Break a third party's terms of service, API terms, or licence conditions.
Exceed or evade a rate limit, quota, concurrency cap or usage tier set by a third party.
Ignore robots.txt or equivalent machine-readable access signals.
Get around a paywall, login wall, CAPTCHA, bot-detection system or IP block.
Rotate accounts, keys, proxies or IP addresses in order to defeat any of the above.

If you need data from a third-party system at volume, get a licence or an API agreement for it. Orbital is not a way around one.

3.3 Scraping and bulk extraction

You may not use Orbital for mass or systematic extraction of data from systems you do not control, including crawling at a scale or frequency a reasonable operator would treat as abusive, harvesting personal information or contact details from public sources for unsolicited outreach, or rebuilding a third party's dataset, index or content for your own use or resale.

Targeted research - an agent reading a page, a document or a public filing to answer a specific question - is a normal use of Orbital and is not restricted by this section.

3.4 Credentials and account access

Grant an agent the narrowest access that lets it do its job. Do not grant broad or administrative access as a convenience.
Do not share one human's credentials between people, between agents, or between workspaces. Do not use an agent to share credentials with anyone.
Do not connect an account you are not authorised to connect, or grant access that the connected system's own terms prohibit you from delegating.
Do not attempt to read, export or extract a credential that Orbital holds for you, or to use a credential granted for one agent, session or purpose for a different one.
Tell us immediately if you believe a credential you granted has been misused. See AI & Agent Terms §5.4 for how credentials are held and revoked.

3.5 Outbound messages and communications

Where an agent sends email, messages, tickets or calendar invitations from an account you connect:

You remain the sender. Marketing, consent, suppression and unsubscribe rules apply to you exactly as they would if you had typed the message.
Do not use agents to send bulk unsolicited commercial messages, to generate volume that would trip a provider's spam controls, or to contact people who have opted out.
Do not use agents to make automated calls or send automated texts without the consents those channels require.

3.6 Do not defeat Orbital's own controls

You may not, and may not instruct an agent to:

Evade or disable a permission scope, approval gate, spend limit, policy check or audit log.
Use one agent to grant another agent authority you did not grant.
Tamper with, delete or falsify an audit record.
Craft prompts or instructions whose purpose is to get an agent to act outside the authority you configured for it.

Governance controls that the customer can talk their way around are not controls. We treat an attempt to defeat them as a serious breach, whoever or whatever made the attempt.

3.7 Do not attack Orbital, other customers, or the model providers

No probing, scanning, load-testing or penetration-testing of Orbital without our prior written permission.
No attempt to access another customer's workspace, data, logs or agents.
No uploading or generating malware, or using Orbital to build or deliver an attack.
No attempt to extract our models, prompts, policies or system instructions, or to reverse-engineer the platform.
No use of Orbital to generate synthetic data for the purpose of training a competing agent-orchestration product.
No use of Orbital in a way that breaches a model provider's or harness provider's own usage policy. Those policies apply to your use through us. See AI & Agent Terms §3.

4. Rate, resource and capacity limits

Agents consume compute, tokens and third-party API calls continuously and without getting tired. Limits are how a shared platform stays usable.

Stay within the limits of your plan, including any limits on concurrent runs, token or inference spend, storage, connector calls, and agent count. The limits that apply to you are those of your plan or order form.
Do not evade a limit by creating extra accounts or workspaces, or by splitting one workload across several.
Do not resell, sublicense or share your capacity with anyone outside your organisation, or run work for a third party on your account, unless your agreement says you may.
Do not run work unrelated to the service - cryptocurrency mining, general-purpose compute, distributed load generation, media transcoding farms, or anything similar.
Do not generate load that degrades the platform for other customers, whether deliberately or through a misconfigured agent loop.

We may throttle, queue or pause activity to protect platform stability or to stop a runaway agent, and we will tell you when we do. What happens when a run reaches a budget or rate ceiling depends on your configuration; test a limit-triggered run before you rely on it.

5. Security research

We welcome good-faith security research. Report what you find to security@withorbital.ai before testing anything that could affect other customers, and do not access, modify or retain anyone else's data. We will not pursue a researcher who follows that path. Details on the Security page.

6. Reporting a problem

Report abuse, harmful content or a suspected violation to report-abuse@withorbital.ai. Tell us what you saw, where, and when. If you are reporting on behalf of someone whose rights have been affected, say so.

7. How we enforce this policy

We take the narrowest action that fixes the problem. Where we can stop a specific agent, revoke a specific connector or halt a specific run rather than suspending your whole workspace, that is what we will do. Agents sit inside production workflows and we would rather not break the rest of your business to deal with one violation.

Depending on what we find, we may: ask you to stop; disable an agent, a connector or a credential grant; require approval for a class of action; throttle or pause activity; suspend users or the workspace; or terminate the agreement.

Notice and the chance to fix it. Where it is safe to do so, we will tell you what we have found and give you a reasonable opportunity to fix it before we suspend anything.

When we act immediately. We will suspend first and explain afterwards where there is a credible risk of harm to a person, an active security threat, unlawful content, a live risk to another customer or to the platform, or a legal obligation to act. We will tell you as soon as we reasonably can.

Investigation. We may investigate suspected violations, including by reviewing the relevant logs and, where necessary and proportionate, content. Access is limited, logged, and governed by AI & Agent Terms §8.

Appeals. If you think we got it wrong, write to report-abuse@withorbital.ai and we will have someone who was not involved in the original decision look at it, and we will respond within 10 business days.

Reporting to authorities. Where a violation involves conduct we are required to report, or that is seriously unlawful, we may report it.

A violation of this policy is a material breach of the Terms of Service.

8. Changes

We may update this policy. If a change materially reduces what you are permitted to do, we will give you at least 30 days' notice by email or in-product before it takes effect, and you can raise it with us during that period. Other changes take effect when posted. We keep the date of the current version at the top of this page and a dated copy of each previous version.

9. Contact

Constructr, Inc., 1484 Pollard Rd, #3006, Los Gatos, CA 95032, USA

Report abuse: report-abuse@withorbital.ai · Legal notices: legal@withorbital.ai · Security: security@withorbital.ai