Last month, Meta announced that it was going to expand its artificial intelligence services worldwide, and the company informed users in Europe that it would use their public information to train its AI services starting June 26.
The ads sent to Facebook and Instagram users in Europe, informing them that their public posts could be used to train their AI services, including its chatbot, prompted privacy concerns and backlash as users think about the next policy change that would be in place.
But for those living in the United States, where online privacy laws are less strict, Meta AI is already using public posts to train its AI. It’s unclear where else Meta might expand the program.
Sign up for The Morning newsletter from the New York Times
Privacy watchdogs have expressed concerns about data use and a lack of details about exactly what Meta will do with people’s information. But Meta says it is complying with privacy laws and that the information it is collecting will make services more relevant to users in a particular region.
Here’s what you need to know about the AI Meta chatbot and how you can opt out of sharing your information.
Chatbot Meta is its answer to ChatGPT.
Meta AI is a smart assistant software powered by artificial intelligence, available on apps including Facebook, WhatsApp and Instagram; it can be used in feeds, chat and search. Like OpenAI’s ChatGPT, Apple’s Siri or Amazon’s Alexa, it is designed to respond to almost any prompt a user gives it.
For example, you could ask: Who is the greatest tennis player of all time?
“The eternal debate!” Meta AI answered that question. “Although opinions may differ, many experts and fans consider Roger Federer, Rafael Nadal and Novak Djokovic to be among the greatest tennis players of all time.”
Meta AI is powered by LLaMA 3, the company’s new and powerful major language model, AI technology that can chat and create images.
The chatbot learns from Instagram and Facebook posts.
The announcement to European users prompted some backlash on Reddit, Tiktok and Twitter, including in the US, where Meta didn’t have to notify users – and so users may not have realized – that it was train their AI with their public jobs.
When asked, the smart assistant said it learned from “massive text datasets” online. The information came from web pages, books, articles and research papers. But part of the dataset also came from social media posts — including Facebook and Instagram posts, Meta AI said, adding that its training came from “anonymized and aggregated” data.
On a page about its AI generation features, Meta said photos and text from public posts on Instagram and Facebook were used to train its AI generation models but not private posts and private messages. User feedback is fair game for the AI features as well.
A spokesperson for Meta — and its chatbot — didn’t specify exactly how the public information was being used other than to “build and improve AI experiences.” It is unclear when Meta began scraping data from US-based users
In the US, unsubscribe by setting your account to private.
For Meta users in the US, there is no way to stop Meta AI from learning from your public social media posts, as there are no privacy laws specific to this.
“While we do not currently have an opt-out feature, we have built in-platform tools that allow people to delete their personal information from conversations with Meta AI across our apps,” Meta said in a statement Friday.
Those using Meta’s apps within the European Union, Britain, the European Economic Area and Switzerland were told they could opt out, according to Meta.
Here’s how to opt out (for those in Europe).
Visit the Meta Privacy Center from your Facebook account, click on “data settings,” and then click “activity outside of Facebook.” Then select “manage your data” and turn off “data sharing,” as well as “AI model training.”
In EU countries, users will also see “GDPR settings”. From there, users can click on “exercise my rights” and submit a request to opt out. Users must also provide a reason for opting out.
On Instagram, users can tap on “settings,” then “about,” and then “privacy policy,” which will lead to information on Meta AI and how to opt out.
Is it legal for Meta AI to use my data?
In Facebook’s legal terms, that company says “if you share a photo on Facebook, you give us permission to store, copy and share it with others.” Depending on your settings, that photo can be used for other Meta products, according to the company.
In Europe, even with the opt-out feature introduced by Meta to comply with privacy laws, watchdog groups are concerned about the sweeping nature of data use.
The European Center for Digital Rights, known as NOYB (None of Your Business), filed complaints in several European countries about Meta’s policy change.
“Meta doesn’t say what it will use the data for, so it could be a simple chatbot, highly aggressive personalized advertising or even a killer drone,” said Max Schrems, chairman and founder of NOYB , in a news release.
c.2024 The New York Times Company