Meta has announced that beginning December 16, all conversations with its AI assistant across Facebook, Instagram, Messenger, and Ray-Ban smart glasses will be used to determine which ads and recommendations appear in the user’s feed. The company will start notifying users about this change from October 7, but it gives no option to opt out of the new ad targeting system. Users who want to ensure that their AI conversations do not affect their advertising must completely stop using Meta AI for all platforms.
Privacy experts say there are unprecedented surveillance capabilities
Emily Bender, a linguist who studies AI, says that Meta’s move is “surveillance masquerading as personalization” with never-before-seen abilities to extract personal information from users. She cautions that the company is playing the role of capitalizing on what she describes as “the illusion of privacy”-the tendency of people to confide in chatbots things they would never post publicly.
The obvious next thing to worry about is whether chatbots will begin to nudge people to give out information, allowing them to be more targeted by advertisements. Bender believes that Meta is walking a risky line when it comes to directly mining the conversations that take place with their AI assistants and then combining that data with all other information that Meta already collects about users. The company emphasizes it wants to be “super transparent” about giving people a heads-up before using this data in new ways, even if users already thought Meta was doing this.
No opt-out option is incompatible with user control assertions
Meta spokesperson Emil Vazquez told Ars Technica that “there is no opt out for this feature,” in contrast to the company’s blog post, which said that users are “in control” of their AI interactions. The only way of avoiding AI-based ad targeting is by stopping the use of Meta AI completely from all platforms. The company compares AI chat recommendations with other recommendations based on liking photos or following pages, but experts say AI conversations could disclose much more personal information than traditional social media conversations.
Meta has had similar privacy issues when it was revealed that users have had their AI conversations appear in the public feeds without their knowledge. A Dutch court has recently ruled that Meta must honor EU users’ decision not to view invasive personalized feeds, meaning that the company must be required to provide chronological feeds not based on profiling within two weeks.
Global impact affects Billions except protected regions
Most Meta users worldwide will be affected by this update, including those who use Meta’s growing range of smart glasses that offer rich sources of data in the form of voice recordings, images, and videos. Only areas that have stringent legislation on data will be exempt from this new targeting mechanism – this includes the European Union, the United Kingdom, and South Korea.
Mark Zuckerberg, CEO, has been clear about the endgame – conversational AI has to pay for itself via ads or subscription services. This announcement marks the first step at scale to become a part of Meta’s core advertising business, with more than one billion monthly users already interacting with Meta AI. Meta’s promotion of AI conversations as part of its advertising ecosystem represents a qualitative leap in the intensification of data collection practices that turn intimate conversations into commercial targetable data.
While the company is presenting this as enhancing the user experience, privacy experts say it has the potential to facilitate unprecedented surveillance efforts and inspire the design of artificial intelligence systems to mine more personal information. With no opt-out for the majority of users, coupled with stronger protections being applied to EU regions, this development only shows the widening disparity between privacy rights in different jurisdictions and Meta’s drive to monetise its AI investments through revenue from advertising.