EU vs Meta: WhatsApp Must Open Up to ChatGPT and Claude

TL;DR — On 9 June 2026 the European Commission ordered Meta to restore free access to the WhatsApp Business API for rival general-purpose AI assistants, giving the company five working days to comply. The order is an antitrust interim measure under Article 102 TFEU, not a Digital Markets Act fine, and it reverses a ban Meta first imposed in October 2025 then replaced with a fee Brussels considered prohibitive. Meta says it will appeal, calling the decision "regulatory overreach." If it ignores the order, the penalty can reach 10% of global turnover.

---

What the Commission actually ordered

On 9 June 2026 the European Commission adopted an interim measures decision requiring Meta to reopen the WhatsApp Business API to third-party general-purpose AI assistants on the same terms that applied before 15 October 2025. In plain terms: free access, restored quickly, while a longer investigation runs in the background.

The headline number is the deadline. Meta has five working days to comply. The measure then stays in force for the duration of the antitrust probe, which has no fixed end date but cannot run indefinitely. If Meta fails to act, it faces fines of up to 10% of its total annual turnover, plus daily periodic penalties for continued non-compliance.

The legal basis is worth getting right, because early summaries muddled it. This is not a Digital Markets Act case. The Commission is acting under classic EU competition law, treating Meta's conduct as a possible abuse of a dominant position under Article 102 of the Treaty on the Functioning of the European Union (and the equivalent Article 54 of the EEA Agreement). Brussels frames the WhatsApp Business API as digital infrastructure that Meta had opened to outsiders and then effectively closed, comparing it to a refusal to supply.

How we got here: a timeline

The dispute did not start in June. It built over roughly eight months, and the sequence matters because it explains why the Commission reached for the unusual step of interim measures.

The October change was not a blanket ban on automation. Structured business bots — the ones that confirm an order, send a shipping update or push an appointment reminder — stayed allowed. What Meta blocked were open-ended assistants whose main job is answering arbitrary questions, the ChatGPT-and-Perplexity-in-your-chat-thread category. Meta's reasoning was twofold: these assistants generated enormous message volume that strained the system, and they did not fit WhatsApp's template-based billing model, so Meta could not monetise the traffic.

The March 2026 revision is the part that turned a policy dispute into an enforcement action. Meta reopened the door but attached a price, charging AI providers per non-template message in the markets where it was legally required to allow them back. According to the Commission, that pricing could have an effect equivalent to the original exclusion, because the costs would be prohibitive for many operators, start-ups and new entrants in particular. In Brussels' reading, a ban replaced by a paywall that nobody can afford is still a ban.

Which AI assistants are affected

This is where the popular framing — "Meta must open up to ChatGPT and Claude" — needs a careful footnote. The Commission's order is about a category, not a named list. It covers third-party general-purpose AI assistants as a class. In its public communications and in most reporting, OpenAI is the provider explicitly identified; Meta's own rebuttal singles out "OpenAI and some of the largest companies in the world."

The assistants that had actually shipped WhatsApp integrations before the ban included OpenAI's ChatGPT, Perplexity and Microsoft Copilot, alongside venture-backed startups such as Luzia and Poke. Anthropic's Claude was not, in the reporting available, named as one of the assistants Meta blocked — so while Claude clearly falls within the protected category, treat any claim that it was specifically running on WhatsApp and got cut off as unconfirmed.

If you are trying to keep track of who builds these assistants and how they stack up, our roundup of the best free ChatGPT alternatives in 2026 is a useful map of the field, and our explainer on Anthropic's path toward an IPO covers where Claude's maker sits commercially.

What changes for users and developers

For end users in the EU, the practical effect is straightforward: the AI assistants that vanished from WhatsApp should become reachable there again, without the assistant's maker being charged a special toll for the privilege. Whether each provider chooses to relaunch its WhatsApp bot is a separate commercial decision, but the regulatory barrier is meant to come down.

For developers and AI companies, the calculus is more interesting.

The deeper point is distribution. WhatsApp has more than two billion users, and being inside that app is a powerful acquisition channel for any consumer AI product. Losing it pushes assistants back toward their own apps and websites; regaining it restores a direct line to a vast audience. That is precisely why the Commission worried about timing — in fast-moving AI markets, a rival shut out for a year may simply lose the race before any final ruling lands.

Why interim measures, and why now

Interim measures are a rarely used tool in EU antitrust. The Commission deployed them here because of the speed argument. As competition chief Teresa Ribera put it, competition can be lost long before a final decision is adopted. The logic is that if rival assistants are excluded throughout a multi-year investigation, winning the case in 2028 or 2029 would be a hollow victory: the market would already have settled.

This fits a broader pattern of EU scrutiny over how the biggest platforms control access to AI distribution and infrastructure, a theme that runs through everything from app-store rules to the financing of the AI labs themselves — see our coverage of the OpenAI IPO in 2026 for the market backdrop. The WhatsApp case is narrower, but the principle is the same: regulators do not want a dominant gatekeeper deciding which AI products get to reach consumers.

Meta's response

Meta has been blunt. It says it will appeal and frames the order as forcing it to give away a paid product. In its statement, the company argued that the Commission has decided some of the world's largest AI firms can use the paid-for WhatsApp Business product for free, describing the outcome as regulatory overreach subsidised by the European companies that do pay.

An appeal does not pause the obligation. Interim measures are immediately binding, so Meta is expected to comply within the five-day window while it challenges the decision through the EU courts — a process that can take years and runs separately from the underlying antitrust investigation.

The bottom line

The verified facts are clear: a 9 June 2026 antitrust interim order under Article 102, a five-working-day deadline, a 10%-of-turnover stick, and a measure that lasts as long as the investigation. The brief's framing holds up, with three refinements worth keeping straight — the legal basis is competition law rather than the DMA; "free access" means rolling back both the October 2025 ban and the March 2026 fee that replaced it; and the order protects a whole category of general-purpose assistants, with OpenAI named explicitly and Claude covered by implication rather than by name. The real prize is distribution to WhatsApp's two billion users, and the real test is whether interim measures can keep AI markets open while the slow machinery of an antitrust case grinds on.