In the midst of an ongoing regulatory battle with the European Privacy Authorities, Meta announcement On Thursday, his virtual assistant powered by AI, Meta AI, is finally launched in the European Union. The Chatbot type tool will be deployed on Meta’s social platform portfolio, but with a set of more limited features compared to what it offers on its national American market.
In addition, Meta has also confirmed in Techcrunch that Meta Ai will arrive in WhatsApp in the United Kingdom, having so far been limited to Facebook, Instagram and Ray-Ban Meta Glasses since its launch in October.
Meta Ai was available in the United States Since 2023serving as an AI assistant capable not only to discuss and answer questions, but to generate images and to create stylistic selfiesamong others Creative wonders. These features are not yet available in the European version.
Last month, a Meta chat version also landed in a handful of countries through the Middle East and Africa. And now, from this week, Meta AI will be deployed in the 27 EU countries, plus 14 additional European countries (and 21 foreign territories), notably Iceland, Norway, Serbia and Switzerland.
In addition to being able to discuss in head with the assistant between the different applications of Meta, Meta Ai will also be available in group cats, although the launch is staggered – with the functionality which should appear first on WhatsApp, in the EU and – as already indicated – the United Kingdom
Refoulement of confidentiality
The launch of the EU of Meta IA marks the latest efforts of the Facebook parent company to disseminate its AI through the block in the face of regulatory concerns concerning the appearance of user data to form AI models.
While the meta was form a On the content generated by users in the United States for years, the technology giant (among others) faced a decline in the EU due to the complete block Confidentiality regulations – including the General Data Protection Regulations (GDPR) – which means that it must have a valid legal basis to process people’s information to train AI models.
Despite these challenges, Meta last in May began to notify regional users From a future change in privacy policy that informed them, he would start using the content from their comments, interactions, status updates, photos and legends for AI training. Business supported That this data processing was necessary for its IA model to reflect “the various languages, geography and cultural references of the inhabitants in Europe”.
However, in June, Meta was forced to Put these plans on the ice After meticulous examination of the Irish Data Protection Commission (DPC), the META data protection regulator in the EU, which raised concerns concerning the way in which Meta stressed the consent of users to process their data.
In short, Meta had implemented an expensive opt-out process (which means that users had to take measures to prevent their data from becoming an AI training fodder) rather than a simple opt-in-with the company based on a legal basis of the GDPR called “legitimate interests”, affirming that its actions were in accordance with the law. However, the DPC did not agree and Meta was forced to rethink its approach.
The company has confirmed to Techcrunch that the AI assistant version that it launches in the EU has not been trained on local users data – hence the reason why it said that it will not notine users or did not require consent, because it affirms that technology has not been trained on their information.
“The model feeding these META AI features was not trained on the first part data of EU users,” said Anna Dack, director of META innovation communications, EMEA, in Techcrunch.
Meta initially faced regulatory concerns similar to the United Kingdom, which (from Brexit) is outside the EU – but which always has a data protection diet based on the GDPR.
Last summer, the Office of the Information Commissioner of the United Kingdom (ICO) asked Meta to take a break from her IA training plans On concerns about how it helped user data. But after the company adjusted its withdrawal processdoing it slightly less expensiveMeta continued to Launch Meta Ai in the United Kingdom – without explicit objection to the British regulator, although the ICO said that “would monitor the situation”.
When asked if Meta’s IA efforts were still trained in British users data, a spokesperson directed Techcrunch to his Advertisement In September, where he said that he would start to train his models on the content of users “in the coming months”, which suggests that he is not yet ready for the public scene.
“Intelligent cat”
For the moment, Meta Ai in the EU will be limited to what the company invoices as a “intelligent cat function” in six European languages: English, French, Spanish, Portuguese, German and Italian.
In the current state of things, the tool is essentially a chatbot cooked in the different Meta applications, including WhatsApp, Instagram, Messenger and Facebook itself. How it works, is to press a small Blue Circle icon to invoke the assistant, and ask him any question that you could ask in a search engine, such as how to perform a task or to find information on a subject.
As indicated above, the functionality will also land in group cats, but this launch is staggered – starting with WhatsApp, before expanding Messenger and Direct Messaging “Soon”.
Users will be able to call the assistant by typing “@metaai”, then asking a question – for example, where dinner, or the best tourist attraction in a given city.
Although Meta says in her advertisement that Meta Ai has “an advanced understanding of what you are looking for”, he says that this does not refer to any kind of personalized suggestions according to user data; Marketing demand is simply in the context of content search more easily and intuitively, he suggests.
However, the company stresses that this launch represents its “first step” in its efforts to bring more AI in Europe and that it plans to “find parity with the United States” over time. What this probably means are more ruptures between Meta – which was critical AI regulations in Europe – and EU regulators.
Techcrunch contacted the DPC to request its response to Meta AI’s announcement for the EU.
“The DPC, as the main author of supervision of Meta, examined Meta AI in recent months with our colleagues supervision authorities across the EU / EEE and we will keep it during the exam while it takes place to users in the coming weeks,” said a spokesperson.