Acceptable Use & AI Policy

Version: v1.3

Last updated: 30 September 2026

This Policy forms part of the Agreement between Whisprr and the Customer. It applies to the Customer, its Authorised Users, integrations and anyone acting through its account.

1. Core rule

Use the Service only for lawful, authorised business communications within the approved Order and configuration. The Customer remains responsible for its data, instructions, offers, services, messages, personnel, recipients and regulatory obligations.

2. Prohibited use

The Customer must not use or assist others to use the Service to:

a. break a law, court order, regulatory requirement, intellectual-property right, privacy right or applicable Meta/WhatsApp or other channel rule;

b. send spam, deceptive marketing, unlawful direct marketing or messages without the required opt-in, lawful basis or recipient choice;

c. impersonate a person, conceal the sender, misrepresent AI as a human where that would mislead, or fabricate endorsements, authority, availability, prices or facts;

d. facilitate fraud, phishing, malware, credential theft, unlawful surveillance, exploitation, harassment, hate, threats, abuse, trafficking or illegal goods or services;

e. generate or distribute child sexual abuse material, non-consensual intimate content or content that unlawfully exploits a child or vulnerable person;

f. defeat access, safety, billing, usage, rate, tenant or channel controls; probe or disrupt the Service without written authorisation; introduce malicious code; or access another tenant's data;

g. reverse engineer non-public logic, extract source code, scrape at abusive scale, or use the Service or output to build or train a competing service, except where law does not permit the restriction;

h. submit information the Customer is not entitled to use, or intentionally expose passwords, payment-card security codes, secrets or unnecessary sensitive information; or

i. use the Service in a manner likely to cause material harm to a person, Whisprr, its providers or the integrity of the channel.

3. AI safeguards and human responsibility

3.1 AI output may be inaccurate, incomplete, outdated, biased or inconsistent. The Customer must test its use case, approve the source knowledge and response boundaries, monitor performance proportionately, and maintain trained people for escalation and human takeover.

3.2 Unless separately assessed, documented and lawfully approved, the Customer must not use the Service or AI output as the sole basis for:

a. emergency or safety-critical decisions;

b. medical diagnosis or treatment, or legal, financial, credit, insurance or other regulated professional advice;

c. employment, education, housing, credit, insurance, healthcare or other decisions producing legal or similarly significant effects; or

d. biometric identification, emotion inference, social scoring, predictive policing or similarly high-risk surveillance.

3.3 The Customer must not represent AI output as guaranteed, professionally verified or uniquely generated. It must disclose AI involvement where required by law, channel policy or the context necessary to avoid deception.

3.4 Conversations do not automatically become approved knowledge. The Customer must deliberately approve and maintain any reusable knowledge source.

4. Messaging, consent and recipient controls

The Customer must:

a. collect and preserve evidence of required opt-in or other lawful permission;

b. honour opt-outs, objections, quiet hours and suppression lists promptly;

c. use approved templates and categories where the channel requires them;

d. clearly identify the responsible business and avoid misleading sender identities;

e. keep frequency, content and targeting reasonable; and

f. maintain an accessible human or alternative route for complaints, urgent matters and failed automation where appropriate.

Whisprr Credits do not include or authorise Meta/WhatsApp channel charges or unlawful messaging.

5. Data protection and restricted data

5.1 The Customer must minimise personal information, keep it accurate, provide required notices and instructions, and limit user access. It must not use the standard Service for children's information, health information, biometric information, criminal-behaviour information or other special personal information unless the use is lawful and Whisprr has approved the safeguards in writing.

5.2 The Customer must promptly notify Whisprr of a suspected compromised account, unlawful instruction, data leak, misdirected communication or material AI safety failure and reasonably cooperate with containment.

6. Fair usage and technical integrity

The Customer must stay within documented plans, rate limits, user and workspace limits, and reasonable technical usage. It may not create duplicate accounts, manipulate message identity or acceptance signals, or otherwise evade a limit or Credit debit. Normal internal classification, retrieval, orchestration, human replies, human takeover, summaries, retries and identity resolution do not independently consume Credits under the v2 commercial model.

7. Enforcement

7.1 Whisprr may investigate credible abuse and require information or remediation. It may block an output, limit a feature or suspend affected access where reasonably necessary for law, safety, security, provider compliance or serious breach.

7.2 Where practicable, Whisprr will give notice, identify the concern and allow a reasonable cure. Immediate action is permitted where delay creates material risk. Enforcement will be proportionate and access restored when the reason is resolved.

7.3 Serious or repeated breach may result in termination under the Terms. Whisprr may preserve and disclose evidence only where lawfully required or permitted.

8. Reporting and changes

Report suspected abuse or an unsafe output to jordan@whisprr.co.za with enough detail to investigate without sending unnecessary personal information.

Whisprr may update this Policy under the change provisions in the Terms. A material new Customer duty will receive reasonable notice and renewed acceptance where required.