Last month, Meta announced its plan to expand its artificial intelligence services globally, notifying European users that it would start using their public information to train its AI services beginning June 26. This announcement has sparked privacy concerns and backlash, especially among Facebook and Instagram users who were informed that their public posts might be used for AI training, including Meta’s chatbot.

In the United States, where online privacy laws are less stringent, Meta AI has already been utilizing public posts for training purposes. The future expansion of this program to other regions remains uncertain.

Data Collection and Privacy Concerns

The announcement about using public posts for AI training has led to backlash, particularly among European users. Meta AI clarified that its training data includes anonymized and aggregated information from various sources, including social media posts. Private posts and messages are not used for training purposes. Meta also stated that user prompts for AI features are included in the training data. However, the company has not provided specific details on how this public information is utilized, other than to “build and improve AI experiences.”

Privacy advocates are c+oncerned about the use of data and the lack of transparency regarding Meta’s intentions. Meta asserts compliance with privacy laws, claiming that data collection aims to enhance the relevance of its services for users in different regions.

Understanding Meta’s AI Chatbot

Meta’s AI chatbot, designed to rival services like OpenAI’s ChatGPT, is a smart assistant powered by artificial intelligence. It is available on platforms such as Facebook, WhatsApp, and Instagram, and can respond to user queries across feeds, chat, and search.

For example, when asked about the greatest tennis player of all time, Meta AI might respond: “The eternal debate! While opinions vary, many experts and fans consider Roger Federer, Rafael Nadal, and Novak Djokovic among the greatest tennis players of all time.”

The chatbot operates using LLaMA 3, Meta’s advanced large language model capable of engaging in conversations and generating images. It learns from a vast dataset that includes social media posts from Facebook and Instagram, alongside other online text sources.

How to Opt Out (For European Users)

European users, including those in the EU, Britain, the European Economic Area, and Switzerland, have an opt-out option. Here’s how to opt out:

1. Facebook: Visit the Meta Privacy Center, navigate to “data settings,” then “off-Facebook activity.” Select “manage your data” and turn off “data sharing” and “AI model training.”
2. Instagram: Go to “settings,” then “about,” and finally “privacy policy,” where you can find information on opting out of Meta AI.

In EU countries, users will also see “GDPR settings.” From there, they can click “exercise my rights” and submit a request to opt out, providing a reason for their decision.

Opt-Out Options in the United States

For U.S. users, there is currently no way to completely stop Meta AI from using public social media posts due to the absence of specific privacy laws. Meta has developed tools within its platforms that allow users to delete their personal information from chats with Meta AI.

Legal Considerations

Meta’s legal terms state that by sharing a photo on Facebook, users grant permission for it to be stored, copied, and shared, potentially for use in other Meta products. In Europe, even with the opt-out feature, privacy groups have expressed concerns about the broad nature of the data usage.

The European Center for Digital Rights (NOYB) has filed complaints in several European countries, highlighting the lack of clarity regarding Meta’s data usage intentions. Max Schrems, the founder of NOYB, criticized the policy, stating that it could lead to various applications, from simple chatbots to aggressive personalized advertising.

Conclusion

Meta’s expansion of AI services has raised significant privacy concerns, particularly regarding the use of public posts for training AI models. While European users have some opt-out options, U.S. users currently have limited control over their data. Privacy advocates continue to push for greater transparency and more stringent data protection measures.