WhatsApp's Nonremovable Meta AI Sparks Regulatory Alarm

Contents

In reality, though, it’s mostly just annoying

It’s probably not the conspiracy you think

Summary

WhatsApp in the EU introduces Meta AI for basic chat and research tasks, facing potential scrutiny from the European Commission.
Users cannot opt out of, fully disable, or otherwise remove the Meta AI chatbot, raising concerns over privacy and consumer choice.
So far, Meta AI lacks advanced features in the EU — it’s not just annoying, it’s also not that great.

WhatsApp users across the EU recently saw a conspicuous green, blue, and purple circle show up near the New Chat button, wholly disagreeing with the popular app’s classic green color scheme. While WhatsApp has offered Meta AI chatbot access in certain regions for many months, the integration has only just begun to roll out to other markets like Europe.

It’s based on the Llama 3.2 general purpose AI model, or GPAI, and is currently limited to generating or improving message text and answering basic online research questions. Despite not yet being fully functional, EU regulators are already looking into whether the implementation “aligns with EU rules,” according to Veronika Cifrová, European Parliament member (via The Standard).

The search bar in the Facebook app has been rebranded to

Facebook and Instagram’s public Meta AI rollouts also began in 2024.

Meta temporarily halted its AI plans in Europe in June 2024, apparently only now reaching an understanding with regulators to ensure it complies with the EU AI Act. The sweeping legislation requires GPAI to engage in transparent practices that protect individuals’ privacy, forcing Meta to revise its chatbot wrapper before letting EU users on board. On top of the months-long delay, specific Meta AI features like memory personalization and text-based image generation are missing from the app’s EU version.

But the still-limited functionality hasn’t stopped EU regulators from taking notice and raising an eyebrow. Meta announced the feature’s launch, but has yet to offer much information about what it actually does. Even the company’s own promises were somewhat vague, with “brainstorming dinner ideas” and “planning a holiday to Canada” serving as two examples of the apparently groundbreaking tasks its LLM chatbot can currently complete.

WhatsApp’s popularity in Europe makes WhatsApp essentially a must-have, or you’ll never be able to message your friends or contact a company’s customer service agents. Uninterested in another feature nobody asked for — complete with a bright, distracting icon — users have flooded social media looking for a way to remove all traces of the Meta chatbot.

A hand holding a smartphone with icons for AI apps on the screen

Here’s the thing: you can’t get rid of it. You can barely even ignore it, thanks to the icon’s bold, eye-grabbing colors. There’s no definitive word on if or when the EU’s version of Meta AI will get any specific features, and the LLM itself certainly doesn’t know. In fact, you can ask “Is Meta AI available on WhatsApp for Android in Portugal?” and get the correct, affirmative response. When asked about availability in the same region on the Windows app, Meta AI assured me it was, but that’s clearly not the case yet.

It’s probably not the conspiracy you think

A phone held in a hand, displaying the WhatsApp Meta AI chat interface.

Meta insists its GPAI compiles data to “identify patterns, like understanding colloquial phrases or local references, not to identify a specific person or their information.” Given that models like Llama are basically super-high-tech pattern recognition algorithms, it only makes sense that user prompts and responses could contribute to the software’s development. It’s akin to individuals’ Alexa voice prompts shuttling back to an Amazon lab somewhere to refine the assistant’s voice recognition abilities; if that kind of stuff didn’t happen, these tools wouldn’t be where they are today.

But that’s not what users are worried about. Meta has repeatedly claimed that its chatbot isn’t digging through private messages. The company that owns Facebook will never convince a majority of consumers of that, and maybe rightfully so (even though modern data harvesting is so effective, spying on encrypted data wouldn’t make sense).

A neon security camera superimposed on an oversaturated image of sunglasses over a stylized blue background

Many users consider the icon off-putting, and criticize Meta AI’s meager feature set and apparent lost ground to competitors. It seems many people want a messaging app that simply does messaging well. Shoehorning in AI-enabled chatbots without providing any choice understandably gets under people’s skin.

That last point could be what draws the ire of regulators. The European Commission regulatory body recently denounced Apple’s EU geo-blocking and forced Apple to enact a set of broad, cross-platform interoperability changes to avoid abusing its commercial power and market share. Generally, the EC and EU AI Act don’t look fondly on companies restricting users’ ability to disable certain, potentially privacy-implicating features.