New York: Meta has acknowledged that its artificial intelligence models have been trained using all publicly shared text and images from Facebook and Instagram users since 2007. This revelation came after initial denials by Melinda Claybaugh, Meta’s global privacy director, during an investigation by local authorities. Under further scrutiny, Claybaugh admitted that Meta’s AI has indeed utilized user data from as far back as 2007.
Senator David Shoebridge of the Green Party has called for a deeper investigation, asserting that Meta has indiscriminately scraped all public posts on Facebook and Instagram unless users actively chose to set their posts to private. This comprehensive data collection practice raises concerns about user consent and privacy.
Meta’s privacy center and blog posts explain that publicly accessible posts and comments from Facebook and Instagram are used to train their generative AI models. These models are employed to enhance features and contribute to the open-source community. Despite this, the company has not disclosed when data collection began or how long it has been ongoing. Meta clarified that only posts set to public will be used for AI training in the future, but data already collected will not be deleted. Users who posted while minors in 2007, unaware of the future use of their data, have been affected by this practice.
Claybaugh assured that Meta does not collect data from users younger than eighteen. In response to Senator Tony Sheldon’s inquiry, Claybaugh confirmed that publicly available images of his children would be removed, although it remains unclear whether similar actions are taken for accounts created by adults while underage.
Also Read | Myanmar Floods Claim 74 Lives as Typhoon Yagi Causes Widespread Devastation
In Europe, where stringent privacy laws are in place, users have the option to exclude their personal data from AI training, and Meta has complied with these regulations. However, users in other regions, including Australia, lack such options. The potential for future changes in policy or opt-out options for Australian and other non-European users remains uncertain.
Senator Shoebridge highlighted that the absence of robust privacy regulations in Australia allows companies like Meta to continue leveraging personal images and videos, including those of children, for profit.
Also Read | US to Prepare “Substantial” Aid Package for Ukraine Amid Ongoing Conflict
FAQs
Has Meta trained its AI on my posts from Facebook and Instagram?
Yes, Meta has used publicly accessible posts and images from Facebook and Instagram to train its AI models since 2007, unless users have explicitly set their posts to private.
Is it possible to prevent Meta from training its AI on my data?
While setting your posts to private will prevent Meta from using future data, previously public posts will remain in their files.
