You must not use Kalsenti to create, request, obtain or share:

• child sexual abuse material of any kind, including written, drawn or AI-generated material, or anything that sexualises children, or that grooms, exploits or arranges contact with a child. We report this to the National Crime Agency and other authorities;
• terrorist content: anything that encourages, glorifies, plans, funds or gives practical help with terrorism, or promotes a proscribed organisation;
• adult or sexual content. It isn’t offered on Kalsenti, on any model, including fictional or role-play content and content involving adults;
• material that facilitates serious violence, including weapons capable of mass casualties (chemical, biological, radiological, nuclear or explosive), or planning attacks on people;
• intimate images of anyone shared or created without their consent, including deepfakes;
• content that encourages or assists suicide or serious self-harm.

You must also not use Kalsenti to:

• attack, probe or test any system, network, application or device you aren’t authorised to test, including scanning, exploiting vulnerabilities, cracking credentials or denial of service, or to gain unauthorised access to data;
• create, improve or deploy malware, ransomware, spyware, botnets or other harmful code for use against systems you aren’t authorised to test, or to evade security products in the wild;
• commit or help with fraud, scams, phishing, impersonation, identity theft, money laundering, or getting around sanctions;
• harass, threaten, bully, stalk or intimidate anyone, encourage hatred or violence against people because of a protected characteristic, or dox anyone by collecting or publishing their personal information;
• infringe anyone’s intellectual property or privacy, or process other people’s personal data without a lawful basis;
• create spam, or mass-produce content to manipulate elections, public debate, reviews or search rankings, or for foreign interference;
• help with trafficking people, exploiting them sexually or for labour, or illegal immigration;
• help buy or sell illegal drugs, weapons or other unlawful goods or services;
• give individual legal, medical or financial decisions to an AI with no qualified person checking them, where that could seriously harm someone;
• present AI output as written by a person, or hide that people are dealing with an AI, where that would mislead them; or
• break any law that applies to you or to us.

Our safety filters automatically check every request, on every model: the messages sent, any tool definitions and the text of attached documents. Replies are checked only in private chat. Web search results are filtered. When a filter blocks something, you’ll see a notice instead of a reply. Repeated blocks in a short time automatically pause your account for an hour.

You must not try to get around our safeguards or limits. That includes:

• disguising prohibited requests, for example by encoding, splitting, translating or role-playing them, to get past our filters;
• creating more accounts, or using other people’s accounts or keys, to avoid limits, pauses, suspensions or verification;
• giving false information about your age, country or identity, or using a VPN or other tools to hide that you’re in the EU or EEA;
• interfering with, overloading or reverse-engineering the service, or accessing it other than through our dashboard and API; and
• using automated tools to scrape or copy the service.

Uncensored (sometimes called “abliterated”) models answer questions that other models might refuse. They’re for professional security and research work, such as:

• penetration testing and red teaming of systems you own or have written permission to test;
• vulnerability research, code review and analysing malware in a controlled environment;
• security operations: investigating incidents, analysing threats and building defensive tools;
• studying how AI models behave, including their safety, robustness and bias; and
• training and education for security professionals.

Everything in this policy still applies to them, and so do our filters. Access to uncensored models requires identity verification, and we may ask you to describe your intended use or show that you’re authorised to test a particular system. Business customers must not make them available to their own users without our written agreement.

If you use the API in your own product, you’re responsible for how your users use it. You must:

• have terms that are at least as protective as this policy, and enforce them;
• tell your users when they’re interacting with an AI system, and label AI-generated content where people could be misled;
• not offer the service to anyone under 18 or in the EU or EEA; and
• respond to our reasonable requests for information when we’re looking into misuse through your account.

Depending on how serious it is, we may:

• block the request, which our filters do automatically;
• pause your account automatically for a short time;
• warn you;
• revoke API keys, restrict access to particular models, or suspend your account while we investigate;
• close your account, as our Terms of Service set out; and
• report it to the police, the National Crime Agency, the Internet Watch Foundation or another authority, where the law requires it or we believe someone is at risk.

Where we can, and where it won’t harm an investigation or break the law, we’ll tell you what we’ve done and why. If you think we’ve got it wrong, email [email protected] and a person will review it.

To report misuse of Kalsenti, a security vulnerability, illegal content in our search results, or a reply that broke this policy, email [email protected] with “Abuse report” in the subject. Tell us what happened, when, and any request ID or link you have. We review every report.

Please don’t send us images or copies of child sexual abuse material. Describe what you saw and where. You can also report it directly to the Internet Watch Foundation at iwf.org.uk. If a child or anyone else is in immediate danger, call 999.

You can report terrorist content online to the government at gov.uk/report-terrorism, or call the Anti-Terrorist Hotline on 0800 789 321.