Meta Aithe artificial intelligence of the company of Mark Zuckerbergwas made available in Europe through the keys on WhatsApp and Instagram and arriving on Facebook and Messenger, opening debates on privacy issues. After blocking the launch of the AI due to the stringent legislation on the protection of personal data present in the territory of the European Union, the Gigante Tech of Menlo Park in the end decided to release the chatbot, but orphaned some features, such as those relating to the generation of images. In the meantime, the DPC (Data Protection Commission) Irish – which substantially corresponds to our privacy guarantor – announced that he wants to continue keep an eye on my destination As the technology will touch all EU users because of some Unresolved questions about the management of European personal data.
Meta Ai and Privacy in EU: a difficult birth
We take a step back and talk about the process, relatively long and somewhat complex, which brought a few weeks ago a destination to land on European soil. Meta has released its features to the in United States In the 2023but paused the launch in Europe after «intense discussions»With the Irish Data Protection Commission or DPC. These discussions have veered on the formation of its artificial intelligence models on European data.
Meta intended to form its large linguistic models, the so -called Llm (Large Language Models) using content from the material shared by adult users on Facebook and Instagram. After the meetings with the Irish authority, Meta chose to interrupt the procedure, however stating that it was a “I pass back for European innovation, competition in the development of AI and further delays in bringing the benefits of the people to people in Europe».
In a post on the company’s company blog, published on 19 March, this is how the company of Mark Zuckerberg commented on the arrival (late) in the EU of Meta Ai:
It took more time than we would have wanted to put our technology in the hands of people in Europe while we continue to navigate in his Complex regulatory systembut we are happy to be finally here.
The features available for consumers in 41 countries in Europe currently include a chatbot Ai able to answer questions from users of the social media platforms Meta, Facebook, Instagram, WhatsApp and Messenger, useful for making brainstorming, planning their activities, obtaining the answer to questions, but not to generate images (functionality that should arrive further). These last features should arrive in the future.
Meta Ai remains “under observation”
The arrival of Meta Ai called the attention of the DPCthe highest European authority in terms of privacy and protection of personal data, which through a spokesman warned:
The DPC, as a leading supervisory authority by goal, has examined a destination for in recent months together with our supervisory authorities throughout the EU/See (European Union/European Economic Spazio, editor’s note) and we will continue to keep the platform under control while it is distributed to users in the coming weeks.
In fact, the Irish authority for the protection of personal data has many doubts pending on the artificial intelligence tool of Meta. Some of these, probably, concern aspects such as the use of prompts (i.e. the text controls given by users), since they could potentially be shared with partners selected by Meta. This, at least, is what is learned by analyzing the terms of use of Meta Ai, which raises doubts about the possible implications that this could have in some rather delicate areas, such as those inherent in business and health.
Just to make you understand what we are talking about, among the many things reported in the terms of use of the AI of Meta, we read:
When information is shared with the AI, AI can keep them and use them to provide more personalized outputs. The user must not share information that does not want to be used and kept by AI, such as account identification, passwords, financial information or other sensitive topics. (…) If AI cannot respond to the request or if our systems determine that a partner could provide a more relevant or updated response, we could ask third parties to give a response. (…) The information we share with third parties can contain personal information if the request includes personal information on someone. Using Ai, the user invites us to share their information with third parties when the service provides more relevant or useful answers.
Waiting for the competent authorities to move on to the “sieve” destination, the recommendation to not send prompts that contain personal information of some kinds remains very valid: and this applies to the artificial intelligence of Meta, but also for those of all the other companies operating in this field. You never know!
