Is WhatsApp banning AI chatbots?
Only one kind. Since January 15, 2026, Meta's terms bar general-purpose AI assistants offered as the product itself on the WhatsApp Business Platform. A business's own AI agent that answers its customers is still allowed, and regulators have forced Meta to reopen access in Europe.
The headlines called it a WhatsApp AI chatbot ban. That is accurate for companies whose product is the AI itself. A dental clinic, a real estate team or an online shop that uses AI to answer its own customers is outside it. If you are planning that second kind of agent, our WhatsApp automation service covers the build, Meta's message fees and the CRM side. This page goes deeper on the rule itself: what Meta's terms say word for word, where general-purpose assistants are still allowed, what Meta charges them, and what a business should do to keep its own agent clearly on the right side of the line.
Everything below comes from Meta's own terms and developer pages, the European Commission and the Italian Competition Authority, each read on October 30, 2026. We are not lawyers, and this is not legal advice.
What exactly do Meta's WhatsApp terms say about AI providers?
Section 4.7 of Meta's WhatsApp Business Platform terms says AI providers may not use the platform to offer AI when the AI is the "primary (rather than incidental or ancillary) functionality". Meta decides what counts as primary, in its sole discretion.
The rule is section 4.7, "AI Providers", of the Meta Terms for WhatsApp Business Platform (last modified September 23, 2026). Read clause by clause, it does five things:
| Part of section 4.7 | What it says | What it means for a business |
|---|---|---|
| Who is an AI Provider | "Providers and developers of artificial intelligence or machine learning technologies", including large language models, generative AI platforms and general-purpose AI assistants, as Meta determines | The companies that make the AI, not the businesses that use it |
| The prohibition | AI Providers may not use the platform, directly or indirectly, to offer those technologies when they are the "primary (rather than incidental or ancillary) functionality" | An assistant offered to the public as the product is out; AI that supports your own service is not the target |
| Solution Provider exception | You may retain an AI Provider as your Solution Provider | You can build your agent on a third-party model or AI platform |
| Data rule | WhatsApp Business Platform data, including anonymous, aggregate or derived forms, may not be used to build, train or improve AI models, except to fine-tune a model "for your exclusive use" | Your chats cannot feed a vendor's general model |
| Country carve-out | Such technologies "may be made available to businesses in certain countries as set forth here", linking to Meta's AI Providers pricing page | Where a law or regulator requires it, general-purpose assistants are allowed back |
The European Commission, when it opened its investigation on December 4, 2025, read the policy the same way: "Businesses may still use AI tools for ancillary or support functions, such as automated customer support offered via WhatsApp." Meta's own pricing page also says service replies can be powered by a third-party AI solution or by people.
Which WhatsApp chatbots are still allowed, and which are not?
Allowed: agents that answer, book and qualify leads for the business that owns the WhatsApp number. Not allowed outside the carve-out countries: a general "ask me anything" assistant offered on WhatsApp as the product, and any use of WhatsApp chats to train a general model.
Is your WhatsApp AI agent allowed?
Meta draws the line by what the AI is for; the model behind it does not matter. These are our readings of section 4.7 for common setups. The last word belongs to Meta.
| Setup | Our reading | Why |
|---|---|---|
| A clinic's agent that answers questions, books appointments and sends reminders | Allowed | The AI supports the clinic's service; it is ancillary |
| A real estate team's agent that qualifies buyers and hands hot leads to an agent | Allowed | Same: customer support and lead handling for the business |
| A shop's agent that tracks orders and handles returns | Allowed | Automated customer support, the Commission's own example |
| Your agent built on GPT, Claude or Gemini, or inside GoHighLevel | Allowed | Section 4.7 lets you retain an AI Provider as your Solution Provider |
| A public chatbot number where anyone can ask anything | Not allowed, except in carve-out countries | The AI is the primary functionality being offered |
| A business that sells "an AI assistant on WhatsApp" as its main product | High risk | If the AI is what the customer pays for, Meta can treat it as primary |
| Exporting chat history to train or improve a vendor's model | Not allowed | Only fine-tuning for your exclusive use is permitted |
The middle ground is a business whose service is itself AI, such as an AI tutor or an AI writing helper offered on WhatsApp. The terms give no list of examples, and Meta decides "in its sole discretion". If that describes you, get legal advice before you build on WhatsApp. For a typical customer-facing agent, see how we scope an AI customer service agent to one business's information.
Where are general-purpose AI assistants still allowed on WhatsApp?
In the European Economic Area, Brazil and Türkiye. Meta's AI Providers page lists Italy from February 16, 2026, 28 other EEA countries and Brazil from March 11, 2026, and Türkiye from October 15, 2026, where Meta says it is legally required to allow them.
Meta's pricing page for AI Providers (updated October 15, 2026) lists the markets and whether Meta charges AI Providers for each service message:
| Market | AI Providers allowed from | Meta's charge per service message |
|---|---|---|
| Italy (+39) | February 16, 2026 | Charged February 16 to May 12, 2026; no charge from May 13, 2026 |
| 28 other EEA countries, from Austria to Sweden | March 11, 2026 | Charged March 11 to May 12, 2026; no charge from May 13, 2026 |
| Brazil (+55) | March 11, 2026 | Charged at the published AI Provider rate |
| Türkiye (+90) | October 15, 2026 | No charge |
| Everywhere else, including the US, UK, UAE, Canada and Australia | Not listed | General-purpose assistants are barred |
Two regulators drove the European part. On December 24, 2025 the Italian Competition Authority (AGCM) ordered Meta to suspend the terms, finding they "completely exclude Meta AI's competitors from the WhatsApp platform". The European Commission had opened its own case on December 4, 2025, covering the EEA except Italy. On June 9, 2026 it imposed interim measures ordering Meta to restore access for third-party general-purpose AI assistants "under the same terms and conditions that were in place before 15 October 2025", which the Commission notes were free of charge, until it reaches a final decision. The investigation itself is still open.
Even in these markets, AI Providers follow the normal WhatsApp rules: service messages "can still only be sent in response to users, during an open 24-hour customer service window", according to the same Meta page.
How much more does Meta charge an AI provider than a normal business?
Between 2.5 and 9 times the normal service rate. In Brazil, the one market still charged, an AI Provider pays $0.0625 per service message against $0.0068 for other businesses. In every market we checked, it equals Meta's marketing rate.
Per service message: normal business vs AI provider
This is our own comparison of two Meta rate sheets downloaded on October 30, 2026: the AI Providers sheet ("effective October 1, 2026") linked from the AI Providers page, and the standard USD list-rate sheet ("effective October 1, 2026") linked from Meta's pricing page. Only Brazil is still being charged at these rates; Italy and the rest of the EEA stopped being charged on May 13, 2026, and Türkiye is not charged.
| Market | Normal business, service message | AI Provider, service message | Multiple | Same market's marketing rate |
|---|---|---|---|---|
| Brazil | $0.0068 | $0.0625 | 9.2x | $0.0625 |
| Spain | $0.0200 | $0.0707 | 3.5x | $0.0707 |
| Rest of Western Europe (e.g. Ireland, Sweden) | $0.0171 | $0.0592 | 3.5x | $0.0592 |
| Netherlands | $0.0500 | $0.1597 | 3.2x | $0.1597 |
| France | $0.0300 | $0.0859 | 2.9x | $0.0859 |
| Italy | $0.0300 | $0.0795 | 2.7x | $0.0795 |
| Germany | $0.0550 | $0.1365 | 2.5x | $0.1365 |
Arithmetic, Brazil row: $0.0625 / $0.0068 = 9.19. A general assistant answering 10,000 messages a month in Brazil would pay $625 to Meta, where a normal business pays $68 for the same count (before the 1,000 free service messages a normal business gets per number). The Commission's June 9 decision described the fee Meta introduced on March 4, 2026 as, at first sight, "in practice equivalent to the previous access ban".
None of this applies to a business running its own agent. Your agent's replies are billed as ordinary service messages; see our WhatsApp message prices by country for the US, UK, UAE and Australia rates.
What should a business do to keep its WhatsApp AI agent compliant?
Keep the agent about your own business, never use chats to train a general model, give customers a clear path to a person, and only message people who opted in. Those four rules come straight from Meta's terms and its Business Messaging Policy.
Section 4.7 is only one of the rules an agent has to meet. WhatsApp's Business Messaging Policy (last updated September 23, 2026) adds the rest. This is the checklist we build to:
- Scope the agent to your business. Its instructions and knowledge base cover your services, prices, hours and policies. Off-topic questions get a polite refusal and a hand-off, so the AI stays "incidental or ancillary".
- Keep chat data out of model training. Use model providers whose business terms exclude training on your inputs, and do not export chat history to improve any shared model. Fine-tuning a model for your exclusive use is the only training section 4.7 allows.
- Offer a human. The policy says automation must come with "prompt, clear, and direct escalation paths", such as an in-chat transfer to a person, phone, email or a support form.
- Automate only inside the 24-hour window. The policy says "You may use automation when responding during the 24-hour window." Outside it, send approved templates.
- Collect opt-in and honour opt-outs and blocks. The policy requires opt-in permission, clear opt-out instructions per message category, and that you "respect all requests (either on or off WhatsApp) by a person to block" you.
- Do not mislead. The policy has no rule that forces you to say the agent is AI, but it bans communications that "confuse, deceive, defraud, mislead, spam, or surprise people". We name the agent as the business's assistant in its first message.
If you run WhatsApp inside GoHighLevel, HighLevel's policy update note says Conversation AI "remains fully compliant with these new terms" and "does not share or train on your WhatsApp data". For agents built in n8n, our guide to building a WhatsApp AI agent with n8n shows where each of these rules sits in the workflow. Whichever route you take, the accounts and data should sit in your name; our policy on who owns the automations and data explains why.
How did Meta's WhatsApp AI policy change over time?
Meta announced the policy on October 15, 2025 and applied it to existing AI providers on January 15, 2026. Under pressure from regulators it reopened Italy, the EEA and Brazil for a fee, made Europe free from May 13, 2026, and added Türkiye on October 15, 2026.
From ban to partial reopening
| Date | Event | Source |
|---|---|---|
| October 15, 2025 | New terms ban third-party general-purpose AI assistants; applies at once to providers new to WhatsApp | European Commission |
| December 4, 2025 | Commission opens an antitrust investigation (EEA except Italy) | European Commission |
| December 24, 2025 | Italy's AGCM orders Meta to suspend the terms | AGCM |
| January 15, 2026 | Terms apply to AI providers already on WhatsApp | Commission; Meta AI Providers page |
| February 16, 2026 | AI Providers allowed in Italy, charged per service message | Meta AI Providers page |
| March 11, 2026 | 28 other EEA countries and Brazil open, charged | Meta AI Providers page |
| May 13, 2026 | Meta stops charging AI Providers in Italy and the EEA | Meta AI Providers page |
| June 9, 2026 | Commission imposes interim measures: free access on pre-October 15, 2025 terms | European Commission |
| October 15, 2026 | Türkiye added, no charge | Meta AI Providers page |
Meta can update the market list at any time, so check the AI Providers page before you rely on a country being open or closed.
If you are weighing channels, Instagram DM automation runs under separate Meta messaging rules, and an AI chat agent on your website is outside WhatsApp's terms altogether.
More questions about the WhatsApp AI chatbot ban
Can I use ChatGPT in my WhatsApp Business chatbot?
Yes, if the model powers an assistant for your own business, such as answering your customers and booking appointments. Section 4.7 lets you retain an AI Provider as your Solution Provider. What you cannot do is offer a general assistant as the product outside the carve-out countries, or let WhatsApp chat data be used to train or improve the provider's models.
Why did Meta ban general-purpose AI chatbots on WhatsApp?
Meta's terms do not give a reason. The European Commission's view, set out in its June 9, 2026 decision, is that the policy left only Meta's own assistant, Meta AI, accessible on WhatsApp, and that this was at first sight an abuse of a dominant position. The investigation is still open, so that finding is preliminary.
Does the AI chatbot ban apply to the WhatsApp Business app?
Section 4.7 sits in Meta's terms for the WhatsApp Business Platform, the API that CRMs and AI agents connect to. The free WhatsApp Business app is a separate product with its own tools, including Meta Business Agent, Meta's own AI agent. If you connect an outside AI to your number, you are on the Platform and section 4.7 applies.
Will my business be charged the AI Provider rate?
Not if you run an agent for your own customers. Meta's AI Providers page says the policy does not change "How or what Meta charges all other businesses" on the platform. Your agent's replies are ordinary service messages: the first 1,000 per number each month are free, and the rest are billed at your customers' country rate.
Is the European Commission decision final?
No. The June 9, 2026 decision is an interim measure that keeps access open while the investigation continues. The Commission says there is no legal deadline for ending an antitrust investigation. Meta must keep access on the pre-October 15, 2025 terms until a final decision, and can face fines of up to 10% of turnover if it does not comply.
Can I fine-tune an AI model on my WhatsApp chats?
Section 4.7 allows fine-tuning a model "for your exclusive use", as long as the data is not used to create, train or improve any other model. In practice that means a private fine-tune under your own account, with a provider whose terms keep your data out of its shared models. Most business agents do not need fine-tuning at all.
What does Meta label AI Provider messages as?
On Meta's AI Providers page, their non-template messages carry the label general_purpose_ai in webhooks, and the Pricing Analytics API reports them under the category AI_BOT. A business's own agent does not get these labels; its replies show as ordinary service messages in billing.
Is Meta AI still available on WhatsApp?
Yes. The Commission's June 9, 2026 press release says that after the October 2025 policy, only Meta's own assistant, Meta AI, remained accessible on WhatsApp. Meta AI is Meta's consumer assistant; it is separate from Meta Business Agent, which businesses set up to answer their own customers.
Sources and further reading
- Meta, Meta Terms for WhatsApp Business Platform, section 4.7, last modified September 23, 2026, checked October 30, 2026.
- Meta for Developers, New pricing policy for AI Providers and its linked rate sheet, updated October 15, 2026, checked October 30, 2026.
- Meta for Developers, Pricing on the WhatsApp Business Platform and the USD list-rate sheet "effective October 1, 2026", checked October 30, 2026.
- European Commission, Commission imposes interim measures on Meta to preserve free access to WhatsApp for rival AI assistants, June 9, 2026, checked October 30, 2026.
- European Commission, Commission opens antitrust investigation into Meta's new policy regarding AI providers' access to WhatsApp, December 4, 2025, checked October 30, 2026.
- Italian Competition Authority, A576, Meta AI: the Authority orders Meta to suspend the terms excluding competing AI chatbots from WhatsApp, December 24, 2025, checked October 30, 2026.
- WhatsApp, WhatsApp Business Messaging Policy, last updated September 23, 2026, checked October 30, 2026.
- HighLevel changelog, WhatsApp policy update: Business Solution Terms and AI use, checked October 30, 2026.
