Regulators are assessing whether WhatsApp’s scale makes it an essential channel for reaching AI users.
Brazil’s competition authority is increasingly examining whether WhatsApp’s near-universal adoption in the country gives Meta* a unique gatekeeping role in the distribution of AI services.
The question has emerged as a central theme in the agency’s investigation of restrictions on external AI providers under WhatsApp’s new Business API terms, even though some market participants said the platform isn't necessarily essential.
After being queried by Brazil's antitrust authority, corporate clients and AI providers widely described WhatsApp as a critical — but not exclusive — interface for reaching users, with its scale and familiarity setting it apart from alternatives. The Tribunal will have to debate whether these advantages, taken together, are sufficient to treat the platform as "indispensable" in Brazil, if the case moves forward.
In March, while ruling on a preventive measure issued by the Superintendence, Councillor Carlos Jacques Vieira Gomes highlighted the significance of WhatsApp’s scale in Brazil — 148 million users, making the country the second‑largest WhatsApp market globally — and framed Meta as the “orchestrator” of a digital ecosystem. Within such an environment, he argued, powerful platforms can identify when competitors threaten their own services, creating incentives for exclusionary conduct.
At the time, lead Councilor Gomes said given that AI solutions distributed through WhatsApp rely heavily on this channel, the proposed policy change risked cutting off a crucial route to users, particularly in a country where zero‑rating practices may push consumers to choose chatbots that operate inside WhatsApp, not outside it, because in‑app use does not consume mobile data.
Dominant but 'complementary'
Across corporate users, WhatsApp is rarely described as the sole channel for customer interaction. Instead, companies stress its role within broader omnichannel strategies — alongside apps, websites and call centers.
Banco do Brasil, for instance, notes that “WhatsApp has contributed to increasing closeness with customers and facilitating quick and accessible interactions,” while emphasizing that the bank can continue operating without it through other channels. Yet it acknowledges that removing the app would reduce convenience and shift customer preferences.
For Bradesco, the absence of WhatsApp would reduce interaction levels, particularly security-related communications.
“Customer behavior and their preference for handling certain topics via WhatsApp lead to the absence of this channel affecting the level of interaction with customers, due to the reduced reach of notifications and transactional communications, especially those related to security,” Bradesco said.
Other sectors point to scale effects. Rappi highlighted the platform’s “unparalleled range of coverage,” while Dasa warns that losing the channel would “create a significant increase in the volume of calls” and force infrastructure expansion.
Bottleneck
For AI providers, the issue is less about functionality and more about access to users.
Zapia and Luzia answered that they have “direct experience” to report that restrictions on WhatsApp access resulted in sharp drops in usage rather than migration to alternative channels.
“Restricting access to WhatsApp leads most users to stop using the service altogether,” Zapia staid, adding that users don't follow services across platforms because their engagement is tied to their everyday messaging environment. Luzia echoed the conclusion: restricting a channel reduces total usage rather than redistributing it. Zenvia attributes this to friction and habit, noting that requiring users to download apps or switch platforms leads to abandonment.
This dynamic reflects a recurring — but not uniform — pattern in the responses: in many cases, users tend to concentrate their activity in a single channel, particularly when WhatsApp is available. As Zenvia observed, “If the company offers service via WhatsApp, users tend to choose it over web chat or the app.”
At the same time, several providers emphasized that usage can vary depending on the context and business model. Nama, for example, characterized WhatsApp as a complementary channel in B2B settings, while acknowledging that competing without messaging platforms in consumer-facing use cases is “significantly harder.”
Competing without WhatsApp and Telegram Competitive asymmetries
When asked by the Brazilian competition authority, most respondents don't categorically rule out competition without WhatsApp or Telegram. Some — like HiPlatform — argue that alternative channels and emerging technologies could sustain competition.
But in practice, the majority of providers describe meaningful constraints.
Four respondents explicitly said it isn't possible to compete effectively without major messaging platforms, while three said it is possible — though two of those qualify their responses by noting limitations depending on the segment or level of reach.
Some reasons keep reappearing: scale, adoption, acquisition cost, engagement and familiarity.
WhatsApp allows instant access to a vast user base without requiring downloads or onboarding, significantly lowering customer acquisition costs. “Distributing via a messaging platform makes it possible to reach users without marketing costs for downloads,” Hi Platform said.
At the same time, the platform benefits from user habits built over years. Rappi said that even when alternatives exist, they don't "have the same reach, capillarity and familiarity of use as WhatsApp.”
This combination creates a structural advantage: a service present on WhatsApp can leverage frictionless distribution and daily engagement, while competitors outside the ecosystem face higher costs and lower visibility.
Competitive asymmetries
Beyond reach, providers also point to differences in access conditions between Meta’s own AI and third-party services.
Several responses describe technical and contractual constraints affecting independent providers, including rate limits, template requirements and a 24‑hour messaging window. In contrast, Meta AI is described as natively integrated into the interface, with fewer restrictions and greater visibility.
Regarding access to "functionalities, Meta AI "is natively integrated into the WhatsApp interface (search bar, automatic suggestions), while third-party providers operate as separate contacts, requiring the user to proactively search for them and initiate the conversation." For "technical restrictions," Business API "imposes limitations on message format, approved templates, and 24-hour conversation windows that do not apply to the integrated Meta AI,” Nama responded.
Pricing has also emerged as a key concern. Interaction, which operates the Poke assistant, said Meta’s per‑message AI surcharge imposed after regulatory intervention generated “economically very substantial” costs, forcing the company to halt marketing efforts in Brazil and deliberately cap growth. According to the company, the charges — applied on top of standard API fees — quickly reached levels that made expansion difficult.
Interaction argued that the financial burden undermines CADE’s interim measures and called on the authority to increase penalties, saying the current daily fine of 250,000 reais ($49,363) is “insufficient to deter” the conduct and allows Meta to maintain the pricing policy while absorbing the cost. However, that's the maximum daily fine applicable by the authority.
Zapia and Luzia go further, arguing that these types of asymmetries — whether technical or economic — create an uneven competitive environment in which independent providers can't match performance, integration or distribution.
Feedback loops
The role of scale in AI markets isn't limited to distribution — it also creates feedback loops that continuously improve the service over time. Respondents broadly describe a dynamic in which more users generate more data, and more data, in turn, enhances performance, personalization and product development.
Inflection AI said broader user reach enables “the use of training data derived from user interactions,” alongside feedback and usage analysis, resulting in “AI models with better performance and greater applicability.” This reflects a core feedback loop: user interactions feed directly into model refinement, which improves output quality and attracts further usage.
Similarly, Nama points to the same circular dynamic, noting that user interactions generate “valuable data to identify query patterns, refine the relevance of results, and calibrate models.” Higher volumes of usage, it added, expand the base for continuous learning — reinforcing the cycle between adoption and improvement.
Interaction describes these mechanisms more explicitly, linking scale to iterative gains in performance. “AI assistants benefit from large volumes of interactions to refine underlying models through user feedback, error correction and continuous learning”, the company said, adding that larger user bases allow providers to “identify edge cases, tune parameters and improve the quality of responses.”
It also highlights how usage data feeds back into product development as a “continuous improvement of the assistant.”
Access to distribution channels such as WhatsApp becomes part of the feedback loop itself. Limiting access does not only reduce reach — it also interrupts the flow of data and feedback that sustains model improvement, as seen in the responses.
The case is still an administrative inquiry and the preventive measure is halted at the moment due to an ongoing judicial court case between Meta and CADE that's under seal. The investigation needs to be turned into an administrative proceeding by the Superintendence for it to be analyzed by the Tribunal.
*banned and designated as extremist in Russia
Source: MLex
