Meta, like other social media companies, has long relied on user data to help drive innovation and profits, and that’s not changing in this new world of generative AI. But that doesn’t mean there won’t be roadblocks.
The parent company of Facebook, Instagram, Threads, and WhatsApp already has been leveraging user posts on the applications to train Meta AI, its chatbot that rivals others like OpenAI’s ChatGPT and Google’s Gemini. In emails to users in the European Union and United Kingdom, Meta noted changes to its privacy policy when alerting that it would begin doing the same in those regions.
Starting June 26, the plan was to use data from users’ posts and interactions with chatbots to train the generative AI chatbot and the models that underpin it, including its Llama large language model (LLM).
The response was concern by users and almost immediate pushback from member countries and regulatory bodies. The Norwegian Consumer Council objected and said it was sending a legal complaint to the EU complaining that the plan violated the EU’s General Data Protection Regulation (GDPR), writing in a statement that “if you do not want your content to be included in the training data, you must request to be opted out. However, this process has been made deliberately cumbersome by using deceptive design patterns and vague wording.”
EU and UK regulators also criticized the plan over privacy concerns.
Insatiable Need for Data
AI models need massive amounts – and a continuous flow – of data to train on, and for companies like Meta, the huge amounts of user information they gather is an attractive source. However, user data is highly protected by the GDPR, which is good for users in the region. However, the United States has no national policy in place similar to the GDPR or the UK’s Data Protection Act (DPA) – relying more on state-by-state regulations – which leaves U.S. users with fewer protections.
In announcing its plans for the EU and UK earlier this month, Stefano Fratta, global engagement director for Meta privacy policy, wrote in a blog post that using EU user “content that people in the EU have chosen to share publicly on Meta’s products and services” is crucial to giving Meta AI a better understanding of the cultures and languages in the region.
“If we don’t train our models on the public content that Europeans share on our services and others, such as public posts or comments, then models and the AI features they power won’t accurately understand important regional languages, cultures or trending topics on social media,” Fratta wrote. “We believe that Europeans will be ill-served by AI models that are not informed by Europe’s rich cultural, social and historical contributions.”
Meta AI is Out of Europe, for Now
However, he updated the post a week later, writing that Meta was delaying the plan in the UK and EU in response to a request by Ireland’s Data Protection Commission (DCP) and pushing back plans to release Meta Ai in the region.
“This is a step backwards for European innovation, competition in AI development and further delays bringing the benefits of AI to people in Europe,” Fratta wrote. “We remain highly confident that our approach complies with European laws and regulations. AI training is not unique to our services, and we’re more transparent than many of our industry counterparts.”
Meta’s international headquarters is in Ireland, so it would have been difficult to rejects the Irish commission’s request to put the plans on hold. In a brief statement, the DCP said it welcomed Meta’s decision, which “followed intensive engagement between the DPC and Meta. The DPC, in co-operation with its fellow EU data protection authorities, will continue to engage with Meta on this issue.”
Meta and Apple May Be Talking
While its plans in the EU and UK have been stymied, at least temporarily, Meta is moving ahead with efforts to make Meta AI more ubiquitous. According to a Wall Street Journal report this weekend, Meta executives have been talking to their Apple counterparts about integrating Meta’s generative AI technology into Apple’s latest – and AI-focused – version of Siri.
Apple earlier this month unveiled a broad range of initiatives for rapidly expanding AI capabilities throughout its product portfolio. One of those was partnering with OpenAI to integrate ChatGPT into Siri. It’s part of Apple’s larger Apple Intelligence suite of features that will debut this year in iOS 18, iPadOS 18, and macOS.
Users with devices running the upcoming operating systems will be able to leverage ChatGPT. According to reports, Apple is considering doing the same with Meta AI and Llama. It would reduce Apple’s reliance on a single generative AI partner. For Meta, a deal would represent a way to quickly expand the reach of its AI technologies.
The WSJ noted that while the talks are ongoing, nothing has been decided and they could still fall through.
