WhatsApp is quietly carving out national exceptions to a sweeping new policy that threatens to reshape how artificial intelligence chatbots operate on the world’s most popular messaging platform. After first backing down in Italy, the Meta-owned service is now exempting Brazil from a global clampdown on third-party, general-purpose AI bots that run on its business API.

The move follows an intervention by Brazil’s competition authority, which ordered WhatsApp to suspend enforcement of its new rules in the country while regulators investigate whether the policy unfairly disadvantages rival AI providers and favors Meta’s own assistant, Meta AI.

Under the policy, WhatsApp is forcing AI developers that use its business API to stop offering general-purpose chatbots on the platform. These are bots that can answer open-ended questions on a wide range of topics, such as ChatGPT-style assistants or news and information bots, rather than narrowly focused customer service tools.

Developers were told they would have a 90-day grace period, starting January 15, to wind down their services. During that window, they must stop responding to user queries on WhatsApp and send notices to users explaining that their chatbots will no longer function on the app. After the grace period, those bots are supposed to go dark for most of WhatsApp’s global user base.

But in a notice sent to AI providers and seen by multiple outlets, Meta has now carved out a major exception: users with Brazilian phone numbers, identified by the +55 country code. For those users, the shutdown order is effectively suspended.

According to the notice, developers “no longer” need to implement the mandatory auto-reply messages or cease responding to queries when interacting with accounts that have a Brazil country code. In practice, that means Brazilian users can continue chatting with third-party AI bots on WhatsApp, even as the same services are being curtailed elsewhere.

WhatsApp has not publicly detailed the change, nor has it responded to questions about whether the exemption could be expanded or reversed. The company has also not clarified how it will handle edge cases, such as users who travel, change numbers, or use virtual numbers registered in different countries.

The policy itself is significant because it targets a fast-growing ecosystem of AI tools that have been using WhatsApp’s business infrastructure as a distribution channel. Over the past two years, a wave of startups and established AI firms have built WhatsApp-based assistants that can summarize documents, translate text, answer questions, or act as personal digital helpers, all through a familiar chat interface.

WhatsApp’s new rules do not affect all automation on the platform. Businesses can still deploy chatbots for customer support, order tracking, and other tightly scoped tasks. What is being squeezed out are the broad, general-purpose AI companions that compete more directly with Meta’s own Meta AI assistant, which the company is aggressively integrating into WhatsApp, Instagram, and Facebook.

That overlap is at the heart of the concerns raised by Brazil’s Administrative Council for Economic Defense, known as CADE. The regulator has opened an investigation into whether Meta’s policy is exclusionary, potentially abusing WhatsApp’s dominant position in messaging to steer users toward Meta AI and away from independent competitors.

Brazil is one of WhatsApp’s largest and most engaged markets, with estimates placing its user base there in the hundreds of millions. The app is deeply embedded in daily life, used for everything from family chats and political organizing to banking and small business operations. Any change to what services can run on WhatsApp has outsized impact in the country, making it a natural flashpoint for regulatory scrutiny.

Meta has rejected the antitrust concerns and framed the policy as a technical necessity. The company argues that its business API was designed for transactional and customer service messaging, not for the heavy, unpredictable traffic patterns generated by AI chatbots that can field open-ended queries around the clock.

In its response to CADE, WhatsApp said that the surge of AI bots on its business platform had put “a strain on our systems that they were not designed to support.” The company also pushed back on the idea that WhatsApp should be treated as a kind of app store or essential gateway for AI services, insisting that AI companies have plenty of other ways to reach users, including mobile app stores, websites, and direct partnerships.

Meta’s argument is that limiting general-purpose AI bots on WhatsApp is about preserving reliability and performance for the core business use cases the API was built to serve. Critics counter that the timing and scope of the policy, coinciding with Meta’s own rollout of Meta AI, make it difficult to separate technical concerns from competitive strategy.

Brazil is not the first jurisdiction to force Meta to soften its stance. In Italy, the national competition authority raised similar questions about whether the new rules could unfairly disadvantage rival AI providers. In response, Meta granted an exemption for Italian users, allowing third-party AI chatbots to continue operating there while the investigation proceeds.

At the European level, the policy has also drawn attention. The European Commission has opened an antitrust probe into Meta’s treatment of AI chatbots on WhatsApp, examining whether the company is leveraging its messaging dominance to tilt the emerging AI assistant market in its favor. That investigation sits alongside broader EU efforts to regulate large digital platforms and AI systems under new competition and technology frameworks.

For AI developers, the patchwork of national exemptions creates a complex operating environment. A single chatbot service might be allowed to function normally for users in Brazil and Italy, while being forced to shut down or radically change behavior for users elsewhere. That fragmentation could push some companies to rethink their reliance on WhatsApp as a primary channel, even in markets where they are still permitted to operate.

The episode also highlights a broader tension in the AI ecosystem: large platforms like Meta, which control critical distribution channels, are simultaneously gatekeepers and competitors. They can set technical and policy rules that shape what kinds of AI services can reach users, while also building their own AI products that benefit from those same rules.

By carving out Brazil and Italy from its chatbot ban, Meta has signaled that it is willing to adjust its global policies under regulatory pressure. But the core direction of travel remains unchanged: a tighter grip on what kinds of AI experiences can live inside WhatsApp, and a preference for steering users toward Meta’s own assistant.

How far regulators in Brazil, Europe, and potentially other regions are prepared to push back will help determine whether WhatsApp evolves into a relatively closed environment dominated by Meta AI, or remains a more open channel where independent AI providers can compete on equal footing.