Circle AI Terms

Circle AI may provide enabled conversation, voice, receptionist, assistant, agent, content, review-assistance, summary, or other AI functionality identified in the purchased service. Availability and data handling depend on the enabled feature. We do not represent that all these functions are included or currently live.

AI can produce inaccurate, incomplete, biased, inappropriate, or non-unique output. The Customer must assess fitness for its intended use and apply appropriate human review, especially before publishing claims, sending sensitive business communications, or acting on a material recommendation. Automated agents require monitoring, scope limits, testing, and an appropriate escalation path. No feature guarantees professional advice, correctness, business performance, ownership, or non-infringement.

Customers must have rights to inputs, knowledge-base documents, website content, and voice or likeness materials. They may not use AI to deceptively impersonate people, manufacture reviews or testimonials, discriminate unlawfully, facilitate fraud, or provide unauthorized legal, medical, or financial advice. Prohibited automated eligibility decisions and other high-risk uses must not be activated simply because an agent can perform them.

Do not submit PHI, sensitive personal information, or confidential information into AI features where the applicable service-provider terms prohibit that processing. At initial launch, approve knowledge-base materials only after checking those restrictions. Ordinary CRM permissions do not automatically carry over to AI. A supported file upload is not approval to upload a confidential contract, customer record, or regulated information. If an inbound conversation unexpectedly includes restricted information, stop or redirect processing under the applicable incident and human-escalation procedure where technically supported.

Clearly identify AI interactions where required, and do not deceive a person about whether they are speaking with an AI system. Outbound AI voice, call recording, transcription, and marketing may each require distinct permission. Account-level acceptance is not recipient-level consent. Honor applicable verbal and written opt-outs and do not allow agents or automations to circumvent suppression.

AI inputs, outputs, prompts, recordings, transcripts, knowledge-base materials, and related metadata may be processed by the verified providers identified in the applicable processing disclosures. The final feature-specific information must accurately state relevant purposes, retention, training or service-improvement uses, restrictions, and choices. No universal promise that data is never used to train any AI system, that it never leaves the United States, or that it is deleted instantly is established by these Terms.

The standard service is not an emergency advisor or a substitute for a qualified professional. Regulated or consequential uses require separate review and may be forbidden by binding provider rules even when otherwise lawful.