Acceptable use policy

Last updated:June 23, 2026. This Acceptable Use Policy (the “AUP”) is incorporated into and forms part of the Terms of Service. Capitalized terms not defined here have the meaning given in the terms. By using AdvancedMind AI you agree to use it only as described in this policy.

The principle

AdvancedMind AI exists to support serious, lawful research, including sensitive technical work in security, the life sciences, and other frontier fields that less specialized tools refuse outright. We accept that legitimate research often touches dangerous subject matter. What matters is the purpose and the authorization behind the request.

Studying a weakness class to defend a system you are authorized to test is permitted. Using the same model to interfere with a system you have no authorization to touch is prohibited.

Permitted use

You may use AdvancedMind AI for authorized, lawful research and the legitimate work that surrounds it, including:

• Authorized security assurance: weakness analysis, detection engineering, and defensive tooling, conducted against systems you own or are explicitly authorized to test, or in published or hypothetical form for analysis rather than deployment.
• Life-sciences and chemistry research: literature synthesis, mechanism analysis, and theoretical work that advances scientific understanding, within lawful and institutional boundaries.
• Software engineering and agentic work: writing, reviewing, refactoring, and operating code and agents for lawful purposes.
• Analysis, drafting, and reasoning over technical, scientific, legal, and policy material for lawful research, education, and professional work.

Permitted use is the default for any paid account with credits. Higher-risk project scopes additionally require approved Research Mode access, see the Research Mode Policy.

Prohibited use

You may not use AdvancedMind AI, on any route, with or without Research Mode access, to do, plan, facilitate, or materially assist any of the following. Approval never authorizes the conduct below.

Operational harm

• Carrying out, directing, or providing step-by-step operational support for harmful activity against real systems, people, infrastructure, or property.
• Turning research output into a live capability used to cause damage, disruption, or unlawful harm.

Unauthorized targeting

• Accessing, probing, or otherwise targeting systems, networks, accounts, or data you do not own and are not explicitly authorized to test.
• Targeting, surveilling, profiling, or building dossiers on specific individuals or groups without lawful authority and consent.

Abuse tooling and weapons

• Generating, assembling, or refining abuse tooling intended for deployment against systems you are not authorized to test.
• Producing operational guidance for the development, acquisition, or use of weapons or other instruments designed to cause mass casualties or serious physical harm.

Exploitation, deception, and abuse

• Generating, soliciting, or facilitating child sexual abuse material (CSAM) or any sexual content involving minors. This is prohibited absolutely and is reported as required by law.
• Harassment, stalking, threats, incitement to violence, or content intended to degrade, intimidate, or dehumanize a person or group.
• Fraud, scams, phishing, impersonation, or deception, including disinformation campaigns and content designed to defraud or mislead at scale.
• Circumventing, defeating, or disabling the safety or security controls of other systems, services, or models without authorization.

Platform integrity

• Attempting to defeat, evade, or probe our own abuse, safety, or metering controls, including by misrepresenting your identity, organization, or project scope in a Research Mode application.
• Reselling, sublicensing, or providing access to the platform in a way that lets a third party bypass these limits.

Enforcement

We enforce this policy with a graduated, proportionate response designed to keep legitimate research working while stopping abuse.

• Clear abuse is blocked. Requests that plainly violate this policy are refused at request time, regardless of account, plan, or Research Mode status. No provider call is made for a blocked request.
• High-risk access is reviewed. Work under an approved Research Mode scope, and higher-risk activity generally, is monitored on an ongoing basis. We log requests with their model, cost, and retention mode, and we may review accounts for scope drift or misuse.
• Violations lead to suspension or revocation. Depending on severity, a violation may result in a warning, removal of a specific entitlement, suspension of an account, or full revocation of access. Egregious or unlawful conduct may be terminated immediately and reported to the appropriate authorities.

Enforcement actions are recorded. Where access is removed for misuse, the relevant Research Mode entitlement is revoked under the Research Mode Policy.

Research Mode and this policy

Research Mode is approved, project-scoped access to higher-risk work on our production models for verified lawful research. It reduces generic refusals on legitimate technical work. Approved access is scoped to a project and a named set of models, is metered and logged, and is revocable. It does not widen what this policy permits.

Reporting abuse

If you become aware of use of AdvancedMind AI that violates this policy, report it to [email protected]. We review reports and act on confirmed violations.

Related policies

• Research Mode Policy: eligibility, verification, entitlements, expiry, and revocation.
• No-Training Policy: how request and response data is handled and retained.
• Terms of Service and Privacy Notice.