Meta plans to make use of social media content posted by European users of its platforms to coach its Llama AI models.
The company's suite of generative AI tools, called AI at Meta, is barely available in select countries, but Meta plans to launch it in Europe soon.
In a blog entryMeta explained that it needs to coach its models on social media content posted by European users of its platforms to make its AI tools more relevant to them. The problem is that European data protection laws strictly protect personal data.
Meta explained: “If we don’t train our models on the general public content that Europeans share on our and other services, resembling public posts or comments, then the models and the AI features they support is not going to give you the chance to properly understand necessary regional languages, cultures or trending topics on social media.”
The article also identified that OpenAI and Google already does this, so Europeans shouldn't have an issue with Meta using their public Facebook and Instagram content to coach its models.
The debate over whether this violates GDPR data protection laws remains to be ongoing, but Meta's updated terms of service will take effect in Europe on June 26, 2024.
none of what you are promoting
The data protection group NOYB (None of Your Business) has filed complaints in eleven European countries and called on the authorities to initiate an urgent procedure to stop this transformation immediately.
NOYB founder Max Schrems said: “Meta is essentially saying it may well 'use any data from any source for any purpose and make it available to anyone on this planet' so long as it's done via 'AI technology.'”
“This is clearly the other of GDPR compliance. 'AI technology' is an especially broad term. Much like 'using your data in databases,' it has no real legal boundary. Meta doesn't say what the info is used for, so it could either be a straightforward chatbot, extremely aggressive personalized promoting, or perhaps a killer drone.”
“Meta also states that user data may be made available to any 'third party' – and meaning anyone on this planet.”
Schrems said the logical approach can be for users to permit Meta to access their data, but Meta as a substitute offers a “hidden and misleading opt-out form.” At least Europeans get an opt-out form. For users in other countries, that's not an option.
Meta says it “doesn’t use people’s private messages to family and friends to coach our AI systems.”
Moreover, in line with Meta, it ought to be allowed unless European users explicitly opt out. If Meta goes ahead with this, even old, dormant or forgotten European Facebook and Instagram accounts will probably be added to the AI training data pile.
