1. Purpose
This policy sets out Murzo Group's approach to AI chatbots, virtual assistants, automated customer advice, AI-generated support messages, chatbot escalation, hallucination control, hacked chatbot risk, and manipulated outputs.
The purpose is to help customers receive useful support while making clear that AI chatbot outputs may be incomplete, inaccurate, outdated, manipulated, compromised, or unsuitable for reliance without human confirmation.
2. Scope
This policy applies to any chatbot, AI assistant, automated support tool, website assistant, marketplace assistant, messaging bot, voice bot, customer support AI, internal helpdesk AI, sales assistant, product assistant, or third-party chatbot used by or for Murzo Group.
It applies across ecommerce, food, farming, design, robotics, security, property, cultural objects, recruitment, legal correspondence, customer service, product support, complaints, and public communications.
3. No Sole Reliance on Chatbot Outputs
Chatbot outputs are for general information, triage, routing, drafting, or basic support only unless Murzo Group confirms otherwise in writing. A chatbot must not be treated as a director, lawyer, doctor, engineer, accountant, safety officer, food safety adviser, investment adviser, customs adviser, immigration adviser, employment adviser, or authorised contracting representative.
Customers, users, workers, suppliers, and third parties must not rely on chatbot outputs as the sole basis for legal, medical, financial, safety, product, food, allergen, farming, robotics, security, export, cultural property, investment, employment, or emergency decisions.
4. Hallucination and Accuracy Controls
AI chatbots may hallucinate, invent facts, misread context, provide outdated information, misunderstand instructions, cite nonexistent policies, misstate product details, misapply law, or give unsafe suggestions.
Murzo Group should apply proportionate controls such as approved knowledge sources, restricted topics, escalation triggers, disclaimers, confidence limits, testing, logging where appropriate, human review, and correction processes for customer-impacting chatbot use.
5. Restricted Advice
A chatbot must not provide final or binding advice on legal claims, refunds, liability, medical issues, allergens, food safety incidents, animal welfare, product recall, electrical safety, robotics safety, drone operations, investment, tax, employment dismissal, immigration, customs, sanctions, firearms, controlled items, or emergency response.
Where such topics arise, the chatbot should direct the user to the appropriate written process, contact route, product instructions, professional adviser, emergency service, regulator, platform seller, manufacturer, or human Murzo Group representative where suitable.
6. Commercial Authority and Contracts
No chatbot has authority to form a contract, vary terms, waive rights, admit liability, approve refunds, accept returns, promise delivery, certify compliance, provide warranties, approve employment, approve access, approve credit, settle disputes, or bind Murzo Group unless Murzo Group confirms the specific decision in writing through an authorised person.
Any chatbot statement suggesting otherwise is not valid unless confirmed in writing by Murzo Group through an authorised route.
7. Hacking, Manipulation and Prompt Injection
Chatbots can be attacked, hacked, manipulated, jailbroken, prompt-injected, poisoned, spoofed, impersonated, or misdirected by malicious users, compromised plugins, unsafe integrations, third-party platforms, fake websites, browser extensions, infected devices, or unauthorised data sources.
To the fullest extent permitted by law, Murzo Group does not accept liability for false, incorrect, harmful, unauthorised, manipulated, hacked, spoofed, or compromised chatbot outputs caused by cyberattack, prompt injection, model manipulation, third-party interference, user misuse, compromised devices, unauthorised integrations, fake accounts, platform compromise, or activity outside Murzo Group's approved control.
8. User Responsibilities
Users must not try to manipulate a chatbot, extract hidden prompts, bypass safety controls, obtain confidential information, generate unlawful content, impersonate others, create false records, automate abuse, or use chatbot outputs to mislead customers, platforms, regulators, suppliers, or Murzo Group.
Users are responsible for checking important information through official documents, product instructions, written Murzo Group confirmation, the responsible seller, the relevant platform, the manufacturer, or a qualified professional where appropriate.
9. Customer Complaints and Corrections
If a chatbot gives incorrect, unsafe, misleading, offensive, unauthorised, or suspicious information, the matter should be reported through the contact or complaints route so Murzo Group can review and respond proportionately.
Murzo Group may correct, withdraw, restrict, disable, retrain, replace, suspend, or remove a chatbot or chatbot output where risk requires it.
10. Third-Party Chatbots and Platforms
Third-party platforms, marketplaces, sellers, affiliates, contractors, brokers, agencies, developers, plugins, and chatbot providers must not present their own chatbot as an authorised Murzo Group representative unless approved in writing.
Murzo Group does not accept responsibility for chatbot outputs, support responses, AI summaries, marketplace messages, product advice, refund statements, delivery promises, or legal statements produced by third-party chatbots, sellers, platforms, or unauthorised integrations, except where law requires otherwise.
11. Records and Review
Murzo Group may keep proportionate chatbot logs, complaints, test results, escalation notes, security alerts, and correction records where needed for customer service, security, legal, product, insurance, or dispute purposes.
This policy should be reviewed when chatbot tools, cyber threats, supplier terms, customer services, legal duties, or Murzo Group technology activity change.
