Amid the ongoing fight for regulatory with European privacy authorities, Meta announced on Thursday that Meta AI, the virtual assistant that powered AI, will eventually be launched in the European Union. Tools like chatbots will be deployed across the Meta social platform portfolio, although they have a more limited feature set compared to those offered in the domestic US market.
Separately, Meta confirmed to TechCrunch that Meta AI will arrive at WhatsApp in the UK.
Meta AI has been available in the US since 2023 and acts as an AI assistant that can not only chat and answer questions, but also generate images and create style selfies. These features are not yet available in the European version.
Last month, the chat-based version of Meta AI also landed in just a single country in the Middle East and Africa. And now, starting this week, Meta AI will be deployed to all 27 EUs as well as 14 more European countries (and 21 overseas territories), including Iceland, Norway, Serbia and Switzerland.
Not only can you chat with your assistants on a variety of Meta apps, but Meta AI will also be available for group chat, but it’s not a surprise release.
Privacy Pushback
The launch of Meta AI’s EU shows Facebook’s parent company’s latest efforts to spread AI to the block in the face of regulatory concerns about tapping user data to train AI models.
While META has trained AI in user-generated content in the US for many years, the tech giant (among others) is facing a pushback in the EU due to BLOC’s comprehensive privacy regulations, including the General Data Protection Regulation (GDPR).
Despite these challenges, last May Meta began notifying local users of upcoming privacy policy changes that inform them of their use of content through comments, interactions, status updates, photos and AI training captions. The company argued that this data processing is necessary to reflect the “diverse language, geography and cultural references of European people.”
However, in June, Meta was forced to ice these plans following scrutiny by the Irish Data Protection Commission (DPC), the EU Meta’s lead data protection regulator.
In short, Meta had implemented a troublesome opt-out process (where users had to take action to prevent them from becoming training feed for their data) claiming that it was compliant with the law by a company that relies on the legal basis of the GDPR known as “legitimate interests.” However, the DPC opposed and Meta was forced to reconsider its approach.
The company confirmed to TechCrunch that the version of AI assistants launched in the EU is not trained on local users’ data. Therefore, they said they would not notify users or ask for consent as they argue that the technology is not trained in information.
“These models powered by meta AI capabilities are not trained in first-party data from EU users,” EMEA, Meta’s innovation communications manager, told TechCrunch.
Meta initially faced similar regulatory concerns in the UK, which lies outside the EU (since Brexit), but still has a GDPR-based data protection regime.
Last summer, the UK Information Commission (ICO) asked Meta to suspend its AI training plan on concerns about how it could help user data. However, after the company coordinated the opt-out process and made it somewhat troubling, Meta continued to launch Meta AI in the UK without explicit objection from UK regulators, but the ICO said it would “monitor the situation.”
When asked if Meta’s AI efforts are still trained on data for users in the UK, the spokesman has begun training models for user content “in the coming months,” suggesting that they are not ready for the publishing phase yet.
“Intelligent chat”
At this time, EU meta AI is limited to what the company charges as an “intelligent chat feature” in six European languages: English, French, Spanish, Portuguese, German and Italian.
As it stands, this tool is basically a chatbot burned into various Meta apps, including WhatsApp, Instagram, Messenger, Facebook itself, and more. How it works is by tapping on the small blue circle icon to summon an assistant, asking questions you might ask in search engines, such as how to perform a task, or finding information about the topic.
As mentioned above, this feature also lands in group chat, but this launch is incredible. I’m starting with WhatsApp before expanding to Messenger and Instagram’s direct messaging “Soon.”
Users can type “@Metaai” and call assistants by asking questions about where to go to dinner or the top tourist attractions in a particular city.
Meta says it has announced that Meta AI has a “highly understanding of what you’re looking for,” but it doesn’t relate to any kind of personalized proposal based on user data. The marketing claim suggests that it simply lies in the context of searching content more easily and intuitively.
However, the company emphasizes that the launch represents a “first step” in an effort to bring more AI to Europe, and that it plans to ultimately “find equality with the US.” What this probably means is the greater conflict between the meta critical of European AI regulations and EU regulators.
TechCrunch contacted the DPC to request that Meta AI be addressed in response to the EU announcement.
“The DPC as META’s primary supervisory authority has been investigating meta AI over the last few months with fellow supervisors across the EU/EEA and will continue to review it as it has been rolled out to users over the next few weeks,” the spokesman said.