Skip to content

Is Meta AI a privacy threat?

  • by
  • 6 min read

Meta’s AI assistant has started appearing across WhatsApp, Facebook, Messenger, and Instagram. On the surface, this doesn’t seem too big; after all, Google is doing it with Bard, and Microsoft has already launched CoPilot for Windows 11. So why does Meta’s introduction of an AI chatbot in its apps raise questions?

It’s an impressive package powered by Meta’s Llama 3 large-language model. It’s launching in 13 countries and includes abilities such as answering search queries and generating images as you type your prompts. However, the company has an extensive history of invading user privacy, whether to link user data across its platforms or for its ad business.

It still has to make money using ads, and given Meta’s track record regarding user privacy and shady practices, an omnipresent AI across all your most used social media networks and messaging apps becomes a daunting presence.

In the News: Ads during paused YouTube videos might soon be a reality

Meta AI’s dance with privacy

Meta also understands that slapping an AI assistant in a messaging app is bound to come with some questions, and as such, has made quite a few claims around the security and privacy of its AI assistant. For example, WhatsApp claims that Meta AI won’t break the app’s end-to-end encryption and that messages with friends and family are off-limits. This is also mentioned in an FAQ section dedicated to the use of AI on the messaging app’s website.

The most important thing to know is that your personal messages with friends and family are off limits. AIs can read what is sent to them, but your personal messages remain end-to-end encrypted, so no one else, including Meta, can see them.

Meta AI boasts an impressive feature list. | Source: Meta

WhatsApp adds that talking with Meta AI does not link your WhatsApp account information on Facebook, Instagram, or any other Meta apps. Chats with AI will be distinctly marked, and users will have to take action to start a conversation by opening a chat or sending a message to the AI.

Meta AI does collect any messages you send to the chatbot. Whatever you send to the AI “may be used to provide you with accurate information or to improve the model,” so WhatsApp urges users to be cautious of what they say to the AI. That said, the model is also trained to limit the possibility of information about people, such as names, from showing up in other chats, and you can always delete your AI messages with a simple command in the conversation.

Meta reiterates that they use automated technology and human moderators to review AI interactions. This can reduce the likelihood of the models’ outputs, including someone’s personal information, and improve model performance, among other things.

Meta AI is available on Facebook, Instagram, Messenger, and WhatsApp. | Source: Meta

The social media giant further explains that its generative AI models are trained on various sources, including “information publicly available online, licensed data and information from Meta’s products and services.” In case you’re still wondering, publicly shared posts from Facebook and Instagram, including photos and text, were used to train these models.

Whether using Meta AI on WhatsApp, Instagram, or Facebook, you’ll see similar warnings and promises across Meta’s platforms. There’s a healthy number of disclosure messages everywhere. Meta is very clear about what training data it has used for the generative AI models you’ll interact with and what information is leaving your phone. But that’s not all.

Can you trust Meta with your data?

Conveniently, Meta has left out any information regarding ads from its generative AI feature-related documentation. For a company that uses ads as its primary source of income, that’s not very reassuring. Generative AI services, whether ChatGPT, Google’s Gemini, or Microsoft’s Bing Chat, aren’t cheap to run. With Meta AI being a ‘free’ product, much like WhatsApp, Instagram, and Facebook, you’re eventually paying for it with your data.

This is reinforced by the fact that there’s no kill switch to disable Meta AI in any app. You can’t disable the chatbot at all. So, if you’re using WhatsApp, Facebook, Messenger, or Instagram in any of the 13 countries where Meta AI has been released, the only way to get rid of it is to stop using the platform altogether.

To make matters worse, Meta AI is reportedly lying to people when asked whether it can be disabled. When asked how to disable Meta AI, the chatbot does spit out a list of instructions, but the settings it mentions don’t exist. This could indicate a toggle we might see in future updates, but for now, you’ll have to live with Meta AI hiding in your search bars.

Photo by mundissima /
There’s no word on the ad policy surrounding Meta AI. | Photo by mundissima /

Meta has had a history of invading user privacy to build ad profiles for highly targeted ads, resulting in multiple crackdowns in the past. The Norwegian Data Protection Authority banned Meta from tracking users for ads in July 2023, on the heels of a ruling from the Court of Justice of the European Union, which found Meta guilty of unlawfully collecting user data without explicit consent.

Hacked verified Facebook accounts have been known to run ads on the platform, and most recently, the company is facing scrutiny for hosting AI-generated NSFW ads despite its strict policies against adult content on its platforms.

Currently, WhatsApp seems safe from Meta AI, given the promises Meta has made concerning how it collects and uses user data. Facebook, Messenger and Instagram are comparatively riskier, especially if you run public-facing accounts on said platforms. However, things can quickly change in the future.

Meta seems to plan to make these features staples in its apps, get users hooked, and then make policy changes to allow the AI to collect more data for advertising, link profiles across apps, or build ad profiles on its users. Regardless, we currently have no information on Meta AI’s role in the social media giant’s ad business.

Meta AI is still in its early days, with a very limited rollout. However, the company is pushing these features globally, and with no way of disabling them, Meta AI will amass a huge user base in no time. It has the potential to turn into the biggest data farm we’ve seen so far, with every user on every Meta platform being a possible data point.

In the News: ICICI Bank’s iMobile Pay App glitch reveals credit card information

Yadullah Abidi

Yadullah Abidi

Yadullah is a Computer Science graduate who writes/edits/shoots/codes all things cybersecurity, gaming, and tech hardware. When he's not, he streams himself racing virtual cars. He's been writing and reporting on tech and cybersecurity with websites like Candid.Technology and MakeUseOf since 2018. You can contact him here: