In January 2026, Meta began enforcing a new policy for AI chatbots on WhatsApp: general-purpose chatbots — all-purpose assistants not tied to a specific business or product — are restricted on the platform. For most businesses, this is not bad news. It clarifies the right path.
What is restricted: general-purpose chatbots
General-purpose means an AI assistant used by the public for anything — like a consumer chatbot across many topics — without a relationship to the business that owns the number. Meta considers this model outside the platform’s direction.
What remains fine: an AI agent for your customers
An AI agent serving your own business customers — answering product questions, looking up order status, and routing requests to your team — remains a core WhatsApp Business use case. This is how serious platforms use AI: as part of a brand’s customer service, with guardrails and a handoff path to humans.
Safe and effective practices
- AI answers from approved sources, such as catalogs, policies, and official FAQs — not unconstrained general knowledge.
- Every action, such as an order, refund, or booking, passes through an audited workflow rather than a free-form model decision.
- A handoff to a human is always available, with full conversation context.
- Use an official number (WABA) and Meta-approved templates for outbound messages.
- Be transparent: customers know they are interacting with your business assistant.
Closing thought
This policy filters out careless chatbot deployments and favors businesses that build AI agents properly: connected to official data, limited to clear actions, and designed with room for humans. In Wazapin, the AI agent answers questions from approved knowledge, looks up order status, requests refunds for your team to approve, and hands off to your team when it cannot answer.
Calculate your team’s WhatsApp costs
Free calculator — enter chat volume to see estimated costs and savings.
