Meta urges the Australian government not to make changes to privacy laws that prevent AI from using personal information taken from Facebook and Instagram posts to train AI, and argues that AI needs to learn how to “discuss the concept of Australia.”
In submission to the Productivity Committee's review of Data and Digital Technology, published this week, Facebook, Instagram and WhatsApp's parent companies advocated for “global policy adjustments” in the pursuit of the Albanese government of privacy reform in the age of AI.
According to Meta, the generated AI model “needs large and diverse data sets” and cannot rely on synthetic data, i.e. data generated by AI alone. The company said that available databases, such as Australian law, are limited to what can provide AI compared to datasets containing personal information.
“The discussion of human culture, arts and emerging trends is not supported by such legislative texts, and discourse conducted in meta-products represents a critical learning of Australian concepts, reality, numbers, and in particular how users of our products are involved.
“This means that authentic, effective learning to ultimately enhance meaningful communication products is best achieved through those discussions and training that includes artefacts themselves.”
Meta is training AI, Llama and Llama in Facebook and Instagram posts that have been published since last year. The company was ordered to stop training data on posting by users of European-based people, and Meta ultimately offered an opt-out option for EU users.
Guardian Australia reported last year that such options are unavailable to Australian users, as European opt-out options “are compatible with very specific legal frames.”
In a submission by the Productivity Committee, Meta said, “We are concerned that recent developments are a step away from international norms in Australia's privacy regime, and are competing with the obligation to contradict broader digital policy goals and promote age and safe experiences online, and to interfere with industry investment in AI in Australia or in key outcomes.”
Hardware Giant Bunnings appealed to the privacy commissioner that it discovered last year against the company's trial of facial recognition technology in select stores, but aimed at Australian privacy laws. The company told the Productivity Committee that it was “committed to protecting the privacy of its customers,” but believed that “all team members deserve to feel safe in the workplace, and that all customers should be able to shop without fear of harm.”
After the newsletter promotion
“Privacy should be taken into consideration in the context of the strict employer liability and the occupyer's legal obligation to maintain a safe place of work and business.”
Woolworths said it supports privacy reforms, but said, “The current structured proposals can pose unnecessary challenges in how to serve customers who increasingly expect personalized engagement and a single, frictionless shopping experience.”
Tech Giant Google said there was too much regulatory uncertainty in Australia around AI, including the proposed AI Guardrails, allowing the company to train AI without violating copyright in search of changes to copyright law.
