Meta to use European social media content to train AI
What's the story
Meta Platforms, the parent company of Facebook, has announced its intention to utilize publicly shared social media content from Europe to train its artificial intelligence models. This decision aligns with Meta's existing practices worldwide, despite previous concerns over Europe's strict privacy and transparency regulations. The company plans to use content from popular platforms such as Instagram and Facebook to train its Llama large language models.
Training process
Meta's AI training methods under scrutiny
Meta has been under scrutiny for its AI training methods, particularly in Europe. The company's top policy executive previously confirmed to Reuters that public Facebook and Instagram posts are used to train its Llama models, with private posts and messages excluded. However, as of April, Meta was still refining its approach to comply with European regulations. The company's chief product officer stated they were "still working on the right way to do this in Europe."
Legal challenges
Meta's transparency efforts face legal challenges
In response to growing concerns over data usage, Meta announced plans to inform Facebook and Instagram users in Europe and the UK about how their public information is used to enhance AI. However, this move has been met with legal challenges from advocacy group NYOB (none of your business). The group argues that Meta's notifications are insufficient under EU privacy rules, which require the company to obtain explicit user consent.