DialLink AI Features Terms

Version 2.0 · Effective November 2, 2026 · Previous versions: Legal Archive

These AI Features Terms are part of the DialLink Terms of Service (the "Terms") and govern your use of AiVA and all other AI Features. Capitalized terms not defined here have the meanings given in the Terms. If these AI Features Terms conflict with the Terms, these AI Features Terms control for the AI Features.

In plain terms (a summary for convenience; the full terms govern): AI Agents are software that speaks and writes on your behalf. You decide what they say, whom they contact and what they are allowed to do in your other systems. They can make mistakes. You must tell Callers they are talking to an AI where the law requires it, obtain the consents the law requires, keep a human reachable, and never use them for emergencies, regulated advice or decisions that seriously affect people's lives. We do not train AI models on your data.

1. The AI Features

1.1 What they are. The AI Features include:

• AI Voice Agents — automated agents that answer or place calls, converse with Callers, answer questions, take messages, book appointments, qualify leads, route or transfer calls and perform actions in connected tools;
• AI Messaging Agents — automated agents that converse with Callers over SMS, WhatsApp or other messaging channels;
• AI Knowledge Bases — content you provide (documents, web pages, FAQs, product information) that AI Agents draw on when responding;
• AI Agent Integrations — connections between AI Agents and third-party systems (for example CRMs, calendars, help desks and payment tools), including through the Model Context Protocol or similar connectors, that allow an AI Agent to read and update data and take actions;
• AI Call Intelligence — transcription, call summaries, action items, tagging, sentiment analysis, suggested replies and similar analysis of calls, voicemails and messages;
• DialLink MCP Server for AI Assistants — a connector that lets a third-party AI assistant you use (for example ChatGPT, Claude or Grok) read your DialLink data and take actions in your Account, such as placing a call or sending a message, within the permissions you grant when you connect it; and
• any Beta Feature that uses artificial intelligence.

1.2 How they work. The AI Features use speech recognition, text-to-speech, speech-to-speech and large language models, including models provided by Model Providers (for example OpenAI and xAI) listed on our Sub-processors page. When an AI Feature runs, the relevant Customer Data (such as call audio, transcripts, messages, your AI Agent instructions and knowledge content, and data from connected tools) is sent to the models to generate Outputs. You choose the model, voice, language and behavior of each AI Agent within the options we offer.

1.3 Not a substitute for human judgment. AI Features assist your business; they do not replace your judgment, your professional obligations or your supervision. Outputs are generated automatically and may be inaccurate, incomplete, out of date, biased or inappropriate, and AI Agents may misunderstand a Caller, mis-transcribe speech (particularly with accents, background noise, names, numbers and specialized terms), give an unintended answer or take an unintended action within the permissions you granted.

2. Your AI Agents Are Your Responsibility

2.1 Configuration. You decide and are responsible for each AI Agent's name, persona, voice, greeting, instructions, knowledge content, language, routing rules, business hours, connected tools and permissions. You must test each AI Agent before putting it into service, review its conversation logs regularly, and correct instructions and knowledge content when Outputs are inaccurate or inappropriate. DialLink does not review your configurations and does not monitor your AI Agents' conversations, except as needed to operate the Services, investigate suspected abuse or respond to your support requests.

2.2 Accuracy, review and reliance. You must independently verify any Output before relying on it or acting on it, including transcripts, summaries, action items, appointment details, lead information, amounts, dates, names and numbers. Transcripts and summaries are not verbatim or complete records of a conversation. Sentiment and intent analysis are estimates. You must not represent Outputs to Callers or third parties as verified, official or guaranteed.

2.3 Human escalation. You must configure every AI Agent so that a Caller can reach a human or leave a message that a human will act on (for example by transfer, callback request or voicemail), and so that the AI Agent clearly tells Callers how to do so when they ask. You must honor a Caller's request to speak with a human, to stop being contacted, or not to be recorded, in accordance with applicable law.

2.4 Connected tools and actions. When you connect an AI Agent to a third-party system, you authorize the AI Agent to act in that system on your behalf with the permissions you grant. You are responsible for granting only the permissions the AI Agent needs, for the security of the credentials you supply, for every action the AI Agent takes in a connected system (including creating, changing or deleting records, sending communications, scheduling and processing transactions), and for reviewing the action logs we make available. We are not responsible for the operation of third-party systems or for actions the AI Agent takes in them, whether or not those actions were intended by you; your remedies for defects in the AI Features are those stated in the Terms.

2.5 Knowledge content. You must have the right to use all content you add to an AI Knowledge Base or AI Agent instructions, and you must keep it accurate. Do not include in instructions or knowledge content any credentials, API keys, payment card numbers, government identifiers, protected health information (unless a Business Associate Agreement is in place), or other data that is not needed for the AI Agent's task.

2.6 Users. You are responsible for how your Users use AI Call Intelligence, including for restricting access to transcripts, summaries and recordings to Users who need it, and for informing Users that their calls may be transcribed and analyzed.

3. Disclosure to Callers and Consent

3.1 AI disclosure. You must configure each AI Agent so that, at the start of an interaction, it tells the Caller that they are speaking with an automated assistant wherever applicable law requires such a disclosure; we recommend doing so in all cases. You control the greeting of each AI Agent and can place the disclosure there. You must never instruct an AI Agent to claim or imply that it is a human, to deny being an AI when asked, or to misrepresent the business it represents. Laws requiring AI disclosure include, among others, the EU Artificial Intelligence Act (for Callers in the European Union), U.S. state laws such as those of Maine and Utah, and sector-specific rules such as California's rules for automated outbound calls; you are responsible for determining which laws apply to your Callers. You must not remove the disclosure for AI Agents that interact with persons in the European Union. We do not monitor or approve your greetings and are not responsible for your decision to omit a disclosure where the law does not require it; we may suspend an AI Agent if we receive a complaint that it is deceiving Callers.

3.2 Recording and transcription notices. AI Features that transcribe or analyze calls involve recording or real-time processing of the conversation. Many jurisdictions, including Florida, California and other U.S. states and most of the European Union, require notice to, or consent from, all parties before a call is recorded, transcribed or monitored. You must give the notices and obtain the consents the law requires — for example through the greeting you configure for your phone system or your AI Agent — wherever it requires them, as described in the Acceptable Use Policy, Section 5.

3.3 Outbound AI calls and messages. If you use an AI Agent to place outbound calls or send outbound messages:

• Calls placed by an AI Agent use an artificial or prerecorded voice for purposes of the Telephone Consumer Protection Act (TCPA) and similar laws. You must obtain the Caller's prior express consent, and for any call or message that includes advertising or telemarketing content, prior express written consent that meets the TCPA's requirements, before the AI Agent contacts them, and you must keep records of consent.
• You must comply with federal and state do-not-call rules, calling-hour restrictions, caller identification and disclosure rules (including rules requiring the calling party's identity to be stated at the beginning of the call and its telephone number during or after the call, and, where required, that the call uses an artificial voice), the requirement that telemarketing calls using an artificial or prerecorded voice offer an automated, interactive opt-out mechanism announced at the start of the call and honored immediately (47 C.F.R. § 64.1200(b)(3)), opt-out and consent-revocation rules, and state telemarketing laws such as the Florida Telephone Solicitation Act.
• You must honor opt-out and consent-revocation requests made by any reasonable means, including replies of STOP, verbal requests during a call, and requests through connected tools, and must stop contacting the Caller immediately and in any event within ten business days as 47 C.F.R. § 64.1200(a)(10) requires.
• You must not use AI Agents for calls or messages that are unsolicited, deceptive, or prohibited by the Messaging Policy in the Acceptable Use Policy.

The liquidated-damages provision in Section 6.4 of the Acceptable Use Policy applies to AI Agent calls and messages.

3.4 Caller data and rights. Callers' personal data processed by AI Agents is Customer Data; you are its controller. You must provide the privacy notices required by law, honor Callers' rights (for example to access or delete their data) using the tools in the Services and the Data Processing Addendum, and collect through AI Agents only the data you need.

4. Prohibited and High-Risk Uses

You must not configure or use an AI Agent or other AI Feature to:

• (a) impersonate a real person (living or dead), including by using a real person's name, likeness or voice without that person's written consent, or clone a voice without the written consent of the person whose voice it is;
• (b) deceive or manipulate Callers, including by denying that the Caller is interacting with an AI, misrepresenting the business or purpose of the call, or exploiting a Caller's vulnerability;
• (c) handle emergencies or safety-critical situations, including calls seeking emergency, medical, police or fire assistance, crisis counseling or suicide prevention; AI Agents must be instructed to direct anyone describing an emergency to hang up and dial 911 (or the local emergency number) and, where you have a human available, to transfer the call immediately;
• (d) deliver advice that constitutes the practice of a licensed profession (for example individualized legal, medical, financial, tax or insurance advice), or diagnose or treat conditions; providing general information, intake, scheduling, triage to a human, and answers drawn from knowledge content you have approved is permitted, provided the AI Agent is instructed to refer questions requiring professional judgment to a licensed person;
• (e) make or materially influence decisions that produce legal or similarly significant effects on a person — including decisions about employment, credit, lending, insurance, housing, education, health care, government benefits or access to essential services — without meaningful human review of each decision, or in a way that unlawfully discriminates against any person, including under laws governing automated decisions such as the Colorado Artificial Intelligence Act and Illinois law on artificial intelligence in employment;
• (f) collect payment card data by voice or message, unless and until we document a payment flow as supported that keeps card data out of recordings and transcripts;
• (g) collect Social Security numbers, government identifiers, biometric identifiers or health information from Callers unless you have a lawful basis, a genuine need and, for health information, a Business Associate Agreement with us; or create voiceprints or otherwise use voice data to identify or authenticate individuals;
• (h) interact with children, or with a service directed to children, without the consents required by law;
• (i) generate or distribute content that is unlawful, defamatory, harassing, threatening, hateful, sexually explicit, or that infringes any person's rights, or that facilitates fraud, scams, phishing, social engineering, malware or illegal activity;
• (j) circumvent safety, content-moderation or rate-limiting features of the AI Features or the Model Providers' models, extract or attempt to reconstruct model weights or training data, or use Outputs to develop a competing model or service;
• (k) violate the usage policies of the Model Providers, as linked on our Sub-processors page and updated by them from time to time, which apply to your use of the models through the Services (we will notify Administrators of changes that materially restrict permitted uses);
• (l) use AI Call Intelligence or any other AI Feature to identify or infer the emotions of your Users or other employees in the workplace where you or they are located in the European Union, or otherwise where the law prohibits it; sentiment analysis in the Services is an estimate of the tone of a conversation for coaching and prioritization and must not be used as the basis for decisions about individuals; or
• (m) violate the Acceptable Use Policy or any applicable law.

We may suspend an AI Agent, an AI Feature or your Account immediately if we reasonably believe this Section is being violated, and we may be required by a Model Provider or carrier to do so.

5. Data and AI Models

5.1 No training on your data. DialLink does not use Customer Data — including call audio, recordings, transcripts, messages, AI Agent instructions, knowledge content or conversation logs — to train, fine-tune or otherwise improve machine learning or artificial intelligence models, whether our own or a third party's, and de-identified data derived from Customer Data is not used for that purpose either. Our agreements with each Model Provider listed on our Sub-processors page prohibit the Model Provider from using Customer Data to train its models.

5.2 Processing by Model Providers. Model Providers process Customer Data as our sub-processors, under written terms that restrict their use of the data to providing the service to us, and are listed on our Sub-processors page. Model Providers may retain data for the limited period stated on that page for abuse monitoring and safety purposes under their terms. Model Providers may be located in the United States; international transfers are governed by the Data Processing Addendum.

5.3 Conversation logs and retention. AI Agent conversation logs, transcripts and summaries are stored in your Account as Customer Data until you delete them or the Account is deleted, as described in Section 10.6 of the Terms. You are responsible for deleting them when your legal obligations require it.

5.4 Service improvement. We may use de-identified and aggregated Usage Data about the AI Features (for example latency, error rates, feature usage and the volume of interactions) to operate, secure and improve the Services. We do not use the content of your communications for that purpose.

5.5 Protected health information. DialLink does not currently offer a Business Associate Agreement. Unless and until one is in place between you and DialLink, you must not use AI Features in connection with protected health information (Terms, Section 10.9). If a Business Associate Agreement is offered in the future, AI Features may be unavailable or restricted as it provides.

5.6 AI assistants you connect through the DialLink MCP Server. When you connect a third-party AI assistant through the DialLink MCP Server for AI Assistants, you instruct us to disclose the data the assistant requests, within the permissions you grant, to the assistant's provider (for example OpenAI, Anthropic or the provider of Grok). For this purpose that provider is not our Model Provider or sub-processor: it processes the data under your agreement with it and its own privacy policy. You are responsible for the permissions you grant, for the instructions you give the assistant, for confirming actions that send data outside the assistant (such as placing a call, sending a message or creating or updating a contact) before they are carried out, and for the assistant's use of your data. The MCP Server returns only the records needed for the request; it has no fields for payment card data, protected health information, government identification numbers or credentials and does not request them, but transcripts, message text and contact notes are returned as recorded and may contain whatever a Caller or User said or typed. The MCP Server is authenticated with OAuth 2.0, cannot move money or change your subscription, and does not read your conversation with the assistant beyond the request it receives. You can disconnect the assistant at any time in its own settings or under Settings → Integrations; we may suspend a connection that is abused or that violates the provider's directory policies or ours. Your use of the MCP Server is also subject to the API Terms of Service.

6. Changes to Models and Features

We may add, remove or replace Model Providers, models, voices and languages, and modify how AI Features work, to improve quality, safety, cost or compliance. Where a change materially changes the behavior of an AI Feature you are paying for, we will give you reasonable advance notice where practicable. Model Providers may change or deprecate their models on their own schedules. Beta AI Features are provided under Section 4.6 of the Terms.

7. Fees

AI Features are billed as described on our pricing page and in your Order, which may include per-minute charges for AI Agent calls, per-segment charges for AI Agent messages, per-minute charges for speech-to-text and text-to-speech, and charges from connected third-party services. Usage is measured by our systems. We may apply usage caps or require prepayment for high-volume AI usage.

8. Outputs and Intellectual Property

8.1 Outputs. As between you and DialLink, and subject to the rights of Callers and third parties, you own the Outputs generated for your Account, and we assign to you any rights we may have in them. Outputs may not be unique: similar or identical Outputs may be generated for other customers, and we grant no exclusivity. You are responsible for Outputs as Customer Data under the Terms.

8.2 Our technology. DialLink and its licensors retain all rights in the AI Features, including system prompts, orchestration logic, agent frameworks, voices, models we develop, and improvements to them. You must not reverse engineer, extract or copy system prompts or model behavior, or use the AI Features to develop a competing product.

9. Compliance With AI Laws

Laws regulating artificial intelligence are developing quickly. As between the parties: (a) DialLink's role is that of the technology provider of the AI Features; DialLink will comply with the obligations that laws such as the EU Artificial Intelligence Act impose on it in that role where they apply to DialLink, and will make available the configuration options and information described in the Documentation to help you use the AI Features responsibly; and (b) you are responsible, as the business that deploys AI Agents toward your Callers and Users, for determining how the laws of each jurisdiction where you operate or where your Callers are located apply to your use, including transparency and disclosure obligations, consumer protection, anti-discrimination, telemarketing, recording and data protection laws, and any required risk assessments, notices, records or human-review processes. We will provide reasonable information about the AI Features to help you meet those obligations.

10. Warranties, Indemnity and Liability

The AI Features are provided "as is" as described in Section 13.3 of the Terms. Without limiting that Section, we do not warrant that Outputs will be accurate, complete, reliable, lawful or fit for any purpose, that AI Agents will follow your instructions in every case, or that any particular model or voice will remain available. Your indemnification obligations in Section 14.1 of the Terms apply to claims arising from your AI Agents and your use of the AI Features, and Section 15 of the Terms limits our liability.

11. Definitions

• "AiVA" means DialLink's AI agent product line (AI Voice Agents, AI Messaging Agents, AI Knowledge Bases and AI Agent Integrations) as described in the Documentation.
• "Model Provider" means a third party whose speech, language or other artificial-intelligence models are used to deliver the AI Features, as listed on our Sub-processors page.
• "Outputs" means transcripts, summaries, responses, messages, analyses and other results generated by the AI Features for your Account (actions taken in connected tools are addressed in Section 2.4).
• Other capitalized terms have the meanings given in the Terms.

---

Questions: legal@diallink.com · DialLink Corp., 1688 Meridian Ave, Suite 700, Miami Beach, FL 33139, United States.