Meta to start training its AI models on public content in the EU

Meet Declaration It is Monday that its AI models will be trained on posts and comments on Facebook and Instagram, before the EU The plan to do it is giving breaks In response to the controlling pressure due to the concern of data privacy. The company will start training AI in the EU this week in EU, it is said. Users’ interaction with Meta AI will also be used to train its models.

The announcement comes after a limited edition Meta AI has been launched in the EU Last month, good after the debut in the United States and other global markets.

When the meta has been Its AI training Over the years, in the US-formed materials, it has faced resistance to the EU due to the strict privacy law, especially General Data Protection Regulation (GDPR) of the block, which requires a clear legal foundation for processing personal data for training AI models.

Meta Back to June 2024 The Irish Data Protection Commission (DPC) will break the plan to start training its AI systems using user data in the EU and the UK after the pushback. The DPC controls the meta in the EU and has been working for the protection authorities throughout the block. In September 2024Meta says it is restarting the effort to train its AI systems using public posts from the UK user base.

Fast forward from today; Meta has announced that it will do it with public posts from its EU user base.

“Last year, we delay training our large language models using public materials when the controllers clarified the legal requirements,” Meta said in his blog post. “We welcome the EDPB’s opinion in December, which confirms our main methods to fulfill our legal obligations. Since then we have been constructively appointed with the IDPC and are expected to continue the full benefits of the generator AI to the people of Europe.”

Starting from this week, users in the EU will start receiving applications and email notifications to explain that users in the EU will start using public data and interaction with Meta Meta AI for training. These notifications include a link to a form that allows users to get out of using their data. Meta says that it will respect all the objections to the new submitting the new submitting forms already obtained.

Meta note that it does not use private messages from users under the age of 18 in the EU or does not use its models to train.

“We believe that we have a responsibility to create AI that is not only available for Europeans, but it is built for them,” said Meta. “This is why it is so important for our generator AI models to train different data so that they can understand the incredible and diverse subtleness and complexities that create European communities, which means hyper-local knowledge and jokes and circuses in our products, from dialect and conversation.”

Meta says it is following examples prescribed by companies such as Google and Openai, both have already used European users’ data to train their AI models.

Meanwhile, the DPC is not fully proceeding by verifying how big language model makers are training their AI services. Last week, the regulator Declaration It was investigating Jai Grock’s training.