Tech giant Meta has suspended plans to train large-scale language models in the European Union based on users' public Facebook and Instagram data, following a request from the Irish Data Protection Commission (DPC).
The request comes amid privacy concerns about the company using users' personal information to train its AI models, but Meta claims that it only uses content that users choose to make public, and does not use private messages or content from accounts of Europeans under the age of 18.
In a blog post published on June 14, Meta said it was “disappointed” by the request and that it was a setback for European innovation as without local information it would only provide people with a second-rate experience, so Meta AI in Europe has been put on hold for the time being.
“We are very confident that our approach complies with European laws and regulations. AI training is not specific to our services and we are more transparent than many of our industry peers,” the blog post said.
During this period, Meta said it will continue to work with the DPC to provide Europeans with the same level of AI innovation as the rest of the world, and will respond to specific requests received from the Information Commissioner's Office ahead of the start of training.
Meta has been making a number of upgrades to its AI products, and recently announced it will be revamping its advertising suite with new generative AI tools, including new messaging tools, which are currently being tested in select territories across Asia Pacific. The announcement was made this morning at the 2024 Cannes Lions Festival.
This language model is currently available to users in Australia, Canada, Ghana, Jamaica, Malawi, New Zealand, Nigeria, Pakistan, Singapore, South Africa, Uganda, Zambia, and Zimbabwe.
