Acceptable Use Policy

Who this covers

You are responsible for how your organization uses Thaliq: its members, agents, tools, knowledge bases, channels and API keys. You are also responsible for what your agents say and do, even when a language model generated it.

We apply this policy to everything that runs on the service. If a rule is not clear, ask us at [email protected] before you act.

Illegal and harmful content

You may not use the service to create, send, store or promote:

• Anything that is illegal where you or your end users are, or that helps someone commit a crime.
• Sexual content involving minors, or any content that sexually exploits or endangers a child. We report it to the authorities.
• Sexual content shared without the consent of the people involved.
• Content that threatens or incites violence, terrorism or self-harm.
• Harassment, hate or discrimination against people because of their race, ethnicity, religion, gender, sexual orientation, disability or other protected traits.
• Instructions or help to make weapons, to traffic in people or goods, or to cause mass harm.

Deception and impersonation

• Do not make an agent impersonate a real person, company or authority without their permission.
• Do not make an agent claim to be human when someone sincerely asks, or when the law requires you to say that it is an AI.
• Do not use agents for fraud, scams, phishing, social engineering or spam.
• Do not use agents for disinformation campaigns, fake reviews or election manipulation.

Responsible use of AI agents

Language models can be wrong. Design your agents so that people stay in control:

• Tell your end users that they are talking to an AI agent. The widget shows an AI notice by default; if you turn it off, you must give an equivalent notice yourself.
• Keep a person in the loop, with approvals, for actions that have legal, financial, medical or safety consequences.
• Do not use agents to make decisions with legal or similarly significant effects on people, such as credit, employment, insurance, housing, education or immigration, without human review.
• Do not present an agent’s output as professional medical, legal or financial advice.
• If your agents talk with people who may be in distress, give them a clear way to reach a person and the emergency resources of their country.
• Do not try to get around the safety measures of the language model or the guardrails of Thaliq to produce content that this policy forbids.

Personal data and privacy

• Process personal data only if you have a lawful basis and have given the notices that the law requires.
• Do not put payment card numbers, passwords or authentication secrets in conversations or knowledge bases.
• Do not use the service to scrape or harvest personal data, to run covert surveillance, or to identify people by their biometric data.
• Send special categories of personal data, such as health data, only if the law allows it and our Data Processing Addendum covers it. [TO CONFIRM: which categories we accept]
• Honor your end users’ requests to access or erase their data; the console and the API have tools for it.

Messaging channels

• Follow the policies of the channel provider; for WhatsApp, the WhatsApp Business policies.
• Get each person’s opt-in before you message them first, and respect opt-outs.
• Do not send unsolicited bulk messages or use purchased contact lists.

Platform integrity

• Do not bypass quotas, rate limits, plan limits or billing.
• Do not create several accounts, or use automated sign-ups, to get more out of the Free plan.
• Do not attack the service or load-test it without our written permission, and do not probe it for weaknesses outside a coordinated disclosure.
• Do not access, or try to access, the data of other organizations.
• Keep your API keys and tokens secret, and use publishable keys only in browsers.
• Do not reverse engineer the service except where the law allows it, and do not resell it unless your plan allows it.

Enforcement

We may investigate suspected violations. We may remove content, block sign-ups, limit use, suspend or terminate accounts, and report to the authorities when the law requires it or someone’s safety is at risk. We give notice when possible, and act immediately when we need to protect people or the service.

If you think we made a mistake, reply to our message or write to [email protected] and we will review it.

Reporting a problem

Report abuse of the service to [email protected]. To report a security vulnerability, write to the security address listed on thaliq.com [TO CONFIRM: dedicated security address]; please do not open public issues and do not test against other customers.

Changes to this policy

We may update this policy. Each update is a new version of the document. For material changes we will notify you by email at least 30 days before they take effect.

Contact

Questions about this policy: [email protected].