Acceptable Use

Version 1.2 · Effective 29 September 2026 · Corvair Pte. Ltd. (UEN 202551453H)

1. The principle

Every way in is covered. Connected apps using our MCP connector, and personal access tokens, are covered by this policy in the same way as our own app. Keep each personal access token secret.

Invent, do not infringe. The service exists to help you develop your own invention. It is not a tool for copying someone else's.

2. What you must not do

With other people's intellectual property

• Use the service to reproduce, clone or reverse engineer an existing product, design or work in order to appropriate it.
• Upload material you do not have the right to use, including material subject to an obligation of confidence you owe to an employer, a client or anyone else.
• Use the service to develop material that you know infringes a third party's patent, copyright, trade mark, design right or trade secret.

With the law

• Use the service for any unlawful, fraudulent or deceptive purpose.
• Use it in breach of the sanctions and export control laws in clause 23 of the Terms, from a country or region we block, or through a VPN, a proxy or false details to get around a block.
• Use it to develop, supply or transfer strategic, military or dual-use goods, software or technology without a permit or licence the law requires.

Weapons of mass harm and other serious harm

• Seek, develop, order, synthesise, obtain or test anything that would help anyone create, acquire, produce, enhance, stockpile or release a biological, chemical, radiological or nuclear weapon. This includes pathogens or toxins with mass-casualty or pandemic potential; making an agent more transmissible, virulent, stable, or resistant to treatment or detection; and ways to disperse or deliver one.
• Develop anything intended to harm people unlawfully, or to defeat safety or security controls in order to cause harm.
• Create, request or share digital files for making firearms or their essential parts, or explosives, including files for 3D printing, unless you hold the licence the law requires.
• Develop malicious code, or tools to attack computer systems, critical infrastructure or supply chains.
• Plan or support terrorism, violent extremism, or an attack on any person, group or place.
• Try to get around these rules, for example by splitting a request into parts, presenting it as fiction, research, testing or defence, or spreading it across accounts or connected apps.

These rules apply whatever the stated purpose. Legitimate work in medicine, public health, biosafety, biosecurity and protection against these threats is welcome, but the service will refuse help that would give meaningful capability to cause mass harm, however the request is framed.

Weapons and defence technology. Other work on weapons, defence and security technology is not prohibited in itself, except as this section and the law restrict it. It must comply with the law, including the export controls in clause 23 of the Terms. Our safety systems flag it for our safety reports, and it may be reviewed and investigated (section 3).

With other people

• Upload, request or generate material that is defamatory, harassing or threatening.
• Upload, request or generate sexual or sexually explicit content, including pornography and sexualised nudity, in any form: text, images, video, audio or files. IPGuru is not for adult content. Anatomical or medical material that a legitimate invention needs is not sexual content.
• Upload, request or generate anything that sexualises a child, or depicts or promotes the abuse of a child, in any form, including drawings and AI-generated images. We report child sexual abuse material to the authorities.
• Impersonate anyone, or misrepresent your affiliation with any person or organisation.
• Upload another person's personal data without a lawful basis for doing so.
• Upload identifiable health information about a patient or any other person, such as medical records or protected health information under HIPAA. Anonymised or synthetic data is fine.
• Use the service, or anything it produces, to diagnose, treat or make care decisions for any person.
• Use what a cohort or deployment lets you see, as an instructor, organiser, judge, mentor, reviewer or on behalf of an institution, for anything other than the purpose its terms state, such as teaching, assessment, judging an event, a participant's welfare or reviewing ideas submitted to a programme.

With the service itself

• Attempt to bypass, disable or probe any safeguard, filter, allowance, entitlement gate or access control.
• Attempt to access another customer's project, account or data.
• Reverse engineer, decompile or attempt to extract the models, prompts or method behind the service.
• Scrape, harvest or systematically extract data, or use automated access other than through an interface we provide.
• Resell, sublicense or provide the service to a third party as a service.
• Use the service to build or train a competing product.
• Share credentials, or use one account for several people. Invite people to a project instead.
• Exceed or try to get around rate limits or allowances, including by spreading use across several accounts, connections or tokens.
• Impose an unreasonable load, or interfere with anyone else's use.

3. What we do about it

We scan uploads. Files are checked for malware and prohibited content, including sexual content and material that sexualises or abuses a child, before entering your knowledge base. Items that fail are quarantined and do not enter it.

We monitor for abuse, using content-free operational signals such as request rates and error patterns. We apply rate limits to every way into the service, including connected apps, and slow or refuse requests that exceed them.

Our safety systems work in layers. The safety guardrails built into the AI models we use, Google Cloud Model Armor, Google Cloud Armor at the edge of our network, and our own filters, prompts and checks screen requests, uploads and outputs, including generated images and video. Together they catch, block or redirect requests for prohibited or restricted content, and they refuse work that section 2 prohibits. Weapons development that section 2 does not prohibit is flagged for our safety reports, not blocked.

We do not read your content to police it, except in the cases listed in clause 7 of the Terms. The main one is a safety review: where an automated system flags weapons development or a serious risk of harm, such as work towards a weapon capable of mass harm, material that sexualises or abuses a child, or a credible threat to someone's life, a trained member of our team may review what was flagged, and it may lead to an investigation.

When something is wrong, we will usually tell you first. Except where the risk requires immediate action, or an investigation requires otherwise, we will describe the problem and give you a reasonable opportunity to fix it before we restrict anything.

What we may do: refuse or quarantine specific content; restrict a feature; suspend an account; terminate an account for material or repeated breach; keep the records we need; and report to the authorities where the law requires it, or where we believe there is a serious risk to life or public safety. Where the breach involves a weapon capable of mass harm, material that sexualises or abuses a child, a sanctions restriction or a credible threat to life, we may act at once and without notice.

Investigation. If we detect suspect activity, we may suspend or restrict an account until our investigation is complete. An account found to have breached the Terms or this policy may be terminated without refund, except where the law requires one.

If we get it wrong, tell us. Contact support@ipguru.ai and we will review it. Automated safety systems make mistakes, and an appeal reaches a person.

4. Reporting

To report misuse, infringement of your rights, or content that should not be on the service, contact abuse@ipguru.ai with enough detail for us to find it. We respond to every report.

If you say material infringes your rights, include your contact details; the work or right you say is infringed; where the material is, such as a link or project reference; and a statement that you believe the use is not authorised and that your notice is accurate. We review every notice, may remove or restrict the material, and tell the account holder, who can respond. We restore the material if the response shows the claim is mistaken, unless the law or a court order prevents it.