You saved the contact, you used it for months, and one day the replies stopped. Then a colleague in Berlin told you it works fine for her. Both of you are right, and the reason is a country code.
Is ChatGPT available on WhatsApp right now?
Only if your WhatsApp number carries an EEA country code. OpenAI restored ChatGPT on WhatsApp across the European Economic Area on 13 July 2026, following an EU antitrust order against Meta. Everywhere else the assistant is still unavailable and no return date has been announced. Your ChatGPT WhatsApp status is decided by the block below.
Status block, verified 2026-08-08
| Region | ChatGPT on WhatsApp | Basis |
|---|---|---|
| EEA (27 EU states plus Iceland, Liechtenstein, Norway) | Available. Restored 13 July 2026 via the verified 1-800-ChatGPT contact | OpenAI's own announcement, following the European Commission's interim measures |
| United States | Not available. No announced return date | Outside the only carve-out in Meta's Business Solution Terms |
| United Kingdom (not in the EEA since Brexit) | Not available. No announced return date | Outside the same carve-out |
| India | Not available. No announced return date | Outside the same carve-out |
| Brazil | No announced availability. Meta's terms do permit AI providers on +55 numbers, but OpenAI's July announcement named the EEA only | Meta's terms carve-out; OpenAI announcement scope |
| Rest of world | Not available. No announced return date | Outside the carve-out |
Re-check this block on any of three triggers: a final European Commission decision in case AT.41034, any OpenAI announcement extending 1-800-ChatGPT beyond the EEA, or 60 days from publication, whichever comes first. Nothing else in this article depends on the table staying true.
Eligibility follows your number, not your location
A VPN will not help you and a holiday in Lisbon will not help you either. Meta's Business Solution Terms, last modified 6 March 2026, put the test on the number itself: general-purpose assistant technologies "may be made available to WhatsApp users who have registered phone numbers with a European Economic Area or Brazil country code." Registered country code, not device location, not IP address.
That is also why the UK sits in the unavailable column despite being in Europe. The EEA is the 27 EU member states plus Iceland, Liechtenstein and Norway. Britain left, so +44 is outside the carve-out.
Why did ChatGPT disappear from WhatsApp in the first place?
Meta changed its own rules. On 15 October 2025 it introduced a policy banning third-party general-purpose AI assistants from the WhatsApp for Business API, leaving Meta AI as the only assistant on the platform. The Terms of Service update took effect on 15 January 2026, and OpenAI and Perplexity both wound their WhatsApp assistants down.
The wording in the current terms is broad. Providers and developers of "large language models, generative artificial intelligence platforms, general-purpose artificial intelligence assistants, or similar technologies as determined by Meta in its sole discretion" are "strictly prohibited from accessing or using the WhatsApp Business Solution" where that technology is "the primary (rather than incidental or ancillary) functionality being made available for use."
Read the qualifier, because it is the half most coverage drops. The bar lands on assistants as the primary functionality. Task-specific business bots survive it, and the terms say so directly: "Notwithstanding the foregoing, you may retain an AI Provider as your Third Party Service Provider." A bot that tracks orders, books appointments, answers support questions or chases an abandoned cart is still permitted, everywhere, and none of what follows in this article changes that. What is barred is turning your WhatsApp presence into a distribution channel for an open-ended assistant.
One further clause matters if you already run an AI-assisted support flow: you may not let Business Solution Data train or improve any AI model, with a narrow exception for fine-tuning a model that is for your exclusive use. The no-code agent walkthrough covers what a compliant, task-scoped setup looks like in practice.
Why did it come back in Europe and nowhere else?
Because a regulator ordered it, and the regulator's jurisdiction has a border. On 9 June 2026 the European Commission adopted interim measures against Meta in case AT.41034, ordering it to reinstate third-party general-purpose AI assistants on the WhatsApp for Business API and to keep them there until the investigation ends. The remedy is EEA-shaped because the underlying finding is.
The Commission's press release is the primary document and it is worth reading in full rather than through recaps, because the sequence is more interesting than the summary.
It was never a clean ban and unban
Almost every secondary account tells this as a two-step story: Meta banned rival assistants, the EU made Meta stop. There was a third step in between, and it is the one that mattered.
The Commission opened its investigation in December 2025, issued a Statement of Objections in February 2026, then a supplementary Statement of Objections in April 2026. In between, on 4 March 2026, Meta revised the policy and accepted third-party general-purpose AI assistants back onto WhatsApp. It just charged them for the privilege. In the Commission's words, Meta "charged a fee which, at first sight, is in practice equivalent to the previous access ban."
Meta's own developer documentation shows what that looked like operationally. Charging for AI providers began on 16 February 2026 in Italy (+39), extended on 11 March 2026 to a list of 29 further countries which is precisely the rest of the EEA, and also to Brazil (+55) on the same date. Then, effective 13 May 2026, Meta "will no longer charge 'AI Providers' for non-template messages delivered to users in certain markets." The European entries drop off the rate card on that date. Brazil does not.
So the fee wall came down in the EEA roughly four weeks before the interim measures decision, after the supplementary Statement of Objections had already telegraphed the order. The Commission still issued the decision, which tells you it wanted the access locked rather than voluntary. The order requires reinstatement "under the same terms and conditions that were in place before 15 October 2025, when such access was notably free of charge for all such AI assistants," maintained until a final decision, and it gave Meta five working days to comply.
Why the remedy stops at the EEA border
The legal basis explains the map. The Commission acted under Article 102 TFEU and Article 54 of the EEA Agreement, using the interim-measures power in Article 8(1) of Regulation 1/2003. Its preliminary finding is that Meta, through WhatsApp, "has at first sight held a dominant position in the EEA-wide market for consumer communication applications since at least January 2023."
An EEA-wide dominance finding produces an EEA-wide remedy. It gives an American, British or Indian user nothing, not because anyone chose to exclude them, but because the Commission has no authority to order anything on their behalf. This was only the second time the Commission has imposed interim measures under Regulation 1/2003 in the regulation's history, which is a fair measure of how unusual the intervention was.
Teresa Ribera, Executive Vice-President for Clean, Just and Competitive Transition, framed the urgency this way: "In rapidly evolving markets, competition can be lost long before a final decision is adopted. This is why these interim measures will remain in place for the duration of the investigation, in order to prevent harm that would be almost impossible to repair."
One number that gets misquoted constantly, so here it is precisely. If Meta contravenes the interim-measures decision, the Commission may fine it up to 10% of "the total turnover in the business year preceding the infringement," and may separately impose daily periodic penalty payments of up to 5% of average daily turnover. That is a ceiling on a penalty for non-compliance, not an automatic fine, and not "10% of global annual revenue" as it is often rendered.
What the European ruling does not give you as a business
It reinstates access for AI providers, not for you. The order is about OpenAI and its peers putting their own assistants back on their own numbers. Whether an ordinary business may now run a general-purpose GPT assistant on its WhatsApp number through the Cloud API is a separate question, and the honest answer today is that it is unresolved.
Here is why it stays unresolved rather than resolving in your favour. The carve-out in Meta's terms is drafted around who the technology is "made available to," and the prohibition itself turns on whether the assistant is "the primary (rather than incidental or ancillary) functionality being made available for use, as determined by Meta in its sole discretion." Sole discretion is not a standard you can plan a deployment against. An EEA business could read the carve-out as permission. Meta could read the same clause the other way on any given Tuesday, and nothing in the interim-measures decision constrains that reading, because the decision addresses AI providers' access rather than business customers' use cases.
If you are building a task-specific bot, none of this touches you and the build versus buy comparison is the more useful read. If you want an open-ended assistant answering as your business on a Meta-issued number, you are betting on an ambiguity. That is the honest framing, and it is exactly why the next section exists.
What you actually wanted: GPT replies on your number, not OpenAI's
Most people searching for this do not want to chat with OpenAI's assistant. They want their own WhatsApp number to reply intelligently. Those are different products, and the second one never depended on the ruling at all.
1-800-ChatGPT cannot do it, even where it works. It is OpenAI's number, OpenAI's contact card, OpenAI's conversation. Your customers do not have it saved. Nothing you send through it goes out under your name, none of it lands in your chat history, and you cannot schedule from it, follow up on it, or route a reply into your CRM. It answers questions for whoever messages it. It does not run your inbox.
The distinction that survives every regulatory change is whose number, whose contacts, whose automation. An assistant on OpenAI's number is a consumer utility whose availability is decided by a Commission timetable and a country code. An assistant on your number is infrastructure you control. If you are in the EEA and reading this thinking availability solved your problem, it solved a different one.
Your number, your contacts, your automation. If you are in the US, the UK, India or anywhere else in the unavailable column, no announcement is coming to fix this for you, and there is no free assistant to fall back on. Connect the WhatsApp number you already use, wire it to the model you already pay for, and skip Meta verification, template approval and per-message fees entirely. Start free at blueticks.co/signup.
How do you get GPT-powered replies on your own WhatsApp number?
Three components, and none of them is the Meta Cloud API. You connect your existing WhatsApp account through WhatsApp Web or a hosted gateway, expose send and read as HTTP calls, and put a model between the inbound message and the outbound reply. Regulatory status is irrelevant to all three.
The connection is own-number, not Cloud API. This matters more than it sounds. The Cloud API needs a Meta-issued or migrated number, business verification and approved templates, and it is the surface all of the policy above applies to. The own-number route drives the account you already have, linked once with a QR scan, which is why the AI provider terms are not the governing document. It is also why you get no Meta guarantees. Details of that trade-off are in the own-number REST API guide.
The model sits in a workflow platform, not in WhatsApp. The practical shape is an inbound webhook, an OpenAI node or HTTP call to your model, and an outbound send. A send is a single POST with the recipient in the path and a flat JSON body, using an Authorization: Bearer header:
POST https://api.blueticks.co/v1/scheduled-messages/+441234567890
Authorization: Bearer bt_live_YOUR_KEY_HERE
Content-Type: application/json
{
"type": "text",
"text": "Thanks, I have your order. Shipping Tuesday."
}
The full field list, the n8n, Make and Zapier equivalents and the expression mapping are already written up in the workflow platform walkthrough. I am not going to rebuild it here. If you would rather skip the workflow layer and let an agent hold the connection directly, that is the WhatsApp API for AI agents route.
What this costs you in risk, stated plainly. Nobody can promise your number will not be banned, and anyone offering that guarantee is overclaiming. WhatsApp's terms prohibit unauthorised bulk and automated messaging, and behaviour is what actually drives enforcement. Replying to people who messaged you first is the low end of the risk curve. Blasting a cold list is the high end, and no vendor changes that. Consent is yours to collect and yours to defend. Pacing, caps and opt-in discipline are build requirements, not disclaimers.
Where Blueticks fits, and what it does not do yet
Blueticks connects your existing WhatsApp number to a hosted engine and exposes it as a REST API and a hosted MCP endpoint, so a model can read chats and send messages as you. The free plan is $0 with three scheduled messages at a time, and there is no Meta verification step because there is no Cloud API involved.
Now the part most vendors would leave out. Inside the product, Claude is the connector that is officially supported today. ChatGPT and Gemini are both marked coming soon. The backend already ships a ChatGPT-shaped tool manifest, so the work is real and in progress, but the in-product connector is not live and I am not going to imply otherwise. If you want a GPT model specifically, today that means routing through the REST API and a workflow platform or your own code rather than a one-click connector. If Claude is acceptable, the MCP integration is the two-minute path.
FAQ
Is ChatGPT on WhatsApp banned? It depends entirely on the country code of your WhatsApp number. See the status block near the top of this article, which is dated and is the only place availability is stated. The underlying rule is that Meta's Business Solution Terms bar general-purpose AI assistants from the WhatsApp for Business API, with a carve-out for numbers registered with an EEA or Brazil country code.
Can I put a ChatGPT bot on my own WhatsApp number? Not through the Meta Cloud API with any confidence. Meta's terms bar general-purpose assistants as primary functionality "as determined by Meta in its sole discretion," and the European ruling addressed AI providers' access rather than business use cases, so that question is genuinely open. You can run GPT-powered replies on your own number over WhatsApp Web or a hosted gateway, which is a different mechanism and not governed by those terms.
Are AI chatbots allowed on WhatsApp at all now? Yes, if they have a job. Task-specific business bots for support, orders, bookings and notifications remain permitted worldwide and were never the target. The terms even provide for retaining an AI provider as your third-party service provider. What is barred is an open-ended general-purpose assistant as the primary function.
Will ChatGPT come back to WhatsApp in the US, UK or India? No return date has been announced for any of them. The European reinstatement came from a Commission order grounded in an EEA-wide dominance finding, which has no reach outside the EEA. Absent a similar intervention by another regulator or a voluntary decision by Meta, there is no mechanism in play that would restore it elsewhere.
What is the difference between using 1-800-ChatGPT and running GPT on my own number? Whose number it is. 1-800-ChatGPT is OpenAI's contact answering whoever messages it, with no connection to your customers, your chat history or your CRM. Your own number sends under your name, to contacts who already have you saved, with scheduling, follow-ups and delivery status you can act on.
Does any of this affect the WhatsApp Business app or ordinary business messaging? No. Everything above concerns the WhatsApp for Business API and general-purpose AI assistants on it. Normal business messaging, the WhatsApp Business app, and per-message Cloud API pricing are unchanged by the AI provider policy.
Notes on sourcing
The regulatory facts here come from the European Commission's own press release for case AT.41034, read through the presscorner API rather than the rendered page, which serves no article text. The policy wording comes from Meta's Business Solution Terms and its AI provider pricing documentation. OpenAI's help centre blocks automated access, so the EEA restoration is sourced to OpenAI's own public announcement and to the Commission decision, and the numeric form of the 1-800-ChatGPT contact has been deliberately left out rather than reproduced from a secondary source.
