This policy covers using talksociety.ai and its demos, and running a voice AI agent we have built for you. It forms part of our terms of service. The short version: an agent we build for you tells people it is an AI, never pretends to be a person, and calls out only on the basis you are responsible for having.
That rule is about the agents we build for customers. Our own calls are not all inbound, and we would rather say so here than imply otherwise: the two "the AI calls you" demos ring the number you type into this site, and we may call a business we have found to introduce ourselves. When we call a business ourselves, our agent says it is an AI in its opening words, and if you ask us not to call again, we never call that number again. The two demo calls open by naming themselves as a Talk Society demo call and say they are an AI if you ask; if you did not ask for the call, the agent apologises and ends it.
Using this website and its demos
- Do not put real personal information into a demo — no customer, patient, candidate or employee details, and nothing confidential. The businesses are invented; use invented details.
- Do not try to get around the site's limits, script the demos, mint sessions automatically, or use them to load-test anything.
- Do not use a demo to generate content that is unlawful, harassing, hateful, sexual, or designed to deceive somebody.
- Do not probe, scan or attempt to break the site, the demo sessions or anything they connect to, and do not attempt to extract the underlying models or prompts.
- Do not book a meeting, request a callback, or give us a phone number or email address that is not yours to give.
Calling and texting people
This is the part that matters most, and it is the part the law is strictest about. If an agent we build makes outbound calls or sends messages:
- You must have a lawful basis to contact each person, including prior express consent where the law requires it for automated or prerecorded calls and for texts.
- Honour opt-outs immediately. A request to stop — spoken on a call, replied to a message, or made any other way — is a request to stop. Keep your own do-not-call list and screen against the applicable ones.
- Call at reasonable hours in the recipient's local time, and within any limits that apply to your industry.
- Identify yourself. The agent says which business it is calling for, and that it is an AI, at the start of the call.
- Use a caller ID you are entitled to use. No spoofing, no misleading numbers.
Where calls are recorded, you are responsible for the notice and consent the law requires in the places you are calling — several U.S. states require every party to agree. Tell us what you need and we will build the disclosure into the call.
What an agent will not be built to do
Some requests we decline, and saying so now saves everybody a conversation later. We will not build an agent that:
- claims to be a human being, or that denies being an AI when it is asked, or that imitates a specific real person's voice or identity without their documented consent;
- deceives people — fake urgency, invented endorsements, pretend regulators or debts, or any form of fraud, phishing or social engineering;
- makes the final decision on something that should be a person's: hiring or firing, credit, insurance, housing, immigration status, a medical diagnosis or treatment, or any other decision with a significant effect on somebody's life;
- gives medical, legal or financial advice as though it were a professional, or handles an emergency instead of directing the caller to the emergency services;
- harasses, intimidates, threatens or repeatedly calls somebody who has asked it to stop;
- discriminates against people on the basis of a protected characteristic, or screens them by one;
- targets children, or is designed to exploit an elderly or vulnerable person's confusion;
- collects more sensitive information than the job needs — health, biometrics, precise location, government identifiers — without a clear basis and the protections that go with it;
- breaks the law, infringes somebody's rights, or breaches the terms of a service it connects to.
Keeping your agent honest
- Tell us the truth about your prices, hours, availability and services — the agent will say what you tell it, to every caller.
- Keep a person available for the calls the agent should hand over, and tell us which those are.
- Look after the credentials and the systems you connect, and tell us promptly if you think either has been compromised.
- Do not resell or sublicense the agent as your own platform unless your proposal says you may.
If something goes wrong
Report abuse, a call that should not have been made, or anything an agent said that it should not have, to hello@talksociety.ai. We investigate, and we would rather hear it from you than from the person it happened to.
Enforcement
If this policy is broken we may limit or suspend a demo, a line or an agent — immediately where there is a risk of harm or a legal breach, and otherwise after telling you and giving you a chance to put it right. Serious or repeated breaches can end an engagement under the terms of service. We will tell you what we did and why.
Contact
Talk Society LLC
1111 Lincoln Road, Suite 500
Miami Beach, FL 33139
hello@talksociety.ai