Description

Australian Facebook users’ posts from as far back as 2007 are being used to train Meta’s artificial intelligence models, the company conceded on Wednesday at a tense Senate inquiry in Canberra. The tech giant was accused of flouting laws and its efforts to protect local users were branded “a joke” by Senators probing the company’s approach to deepfakes, algorithms, online safety and tax. Meta representatives confirmed Australian users have no blanket opt out option for AI training because there is no legal requirement for one, unlike in Europe, and that users aren’t reading its 20,000 word privacy policy that details the AI training.

Summary

Meta has acknowledged using Australian Facebook users' posts since 2007 to train its AI models, revealed during a Senate inquiry in Canberra. The company faced criticism for perceived legal and ethical oversights, especially as Australian users lack opt-out options available in other regions like Europe. This practice highlights ongoing concerns about the protection and privacy of user data within AI development and the importance of strict governance in mitigating risks related to AI deployments and potential misuse. The inquiry underscores the need for more robust AI safety and privacy policies, particularly in the context of international tech giants operating in Australia.

Body