Cohere Usage Policy

Cohere's Usage Policy sets out the rules developers and businesses must follow when using Cohere's AI models and API. The single most important thing for developers to know is that they must describe and get approval for their specific use case before accessing the API, and they are personally responsible if their applications cause harm — including any harm caused by their own end users. If you are building a product with Cohere's technology, review the prohibited use categories carefully, as violations can result in termination of access.

This document is Cohere's Usage Policy (Acceptable Use Policy) governing access to and use of the Cohere API and related AI models, with legal basis established through contractual agreement at the point of API access. The most significant obligations created include a mandatory use-case disclosure and approval requirement prior to API access, downstream operator accountability for ensuring end-user compliance, and affirmative prohibitions on specific categories of harmful outputs. Notable provisions that deviate from industry standard include an explicit prohibition on generating 'Non-Consensual Intimate Imagery (NCII),' a broad restriction on AI-generated content for 'influence operations' and 'disinformation,' and a requirement that developers document potential harms of their applications — an affirmative risk-assessment obligation uncommon in standard API ToS. The policy engages the EU AI Act (particularly high-risk AI system provisions), FTC Act Section 5 on unfair or deceptive practices, and GDPR/CCPA through its privacy violation prohibitions; compliance teams should note that the policy's prohibition on 'surveillance' and 'tracking individuals without their knowledge' may interact with obligations under ECPA and state wiretapping statutes, and the broad 'political content' restriction warrants review for platforms operating in regulated electoral or political advertising contexts.