Meta has warned the Australian Government that its proposed privacy reforms could impede innovation and hamper the development of artificial intelligence (AI) models, as major companies including Google, Amazon, Woolworths and Telstra respond to a wide-ranging Productivity Commission inquiry into the future of data regulation.
The submission forms part of the Productivity Commission’s Pillar 3: Harnessing Data and Digital Technology consultation, one of five key “productivity pillars” identified by the Commission to lift long-term economic growth and resilience.
The Pillar 3 review explores how smarter use of data and digital tech can drive national productivity, asking for feedback on four areas of reform: modernising privacy laws, enhancing consumer data access, encouraging digital financial reporting and enabling the safe, scalable use of AI.
At the heart of the consultation is a critical question: How can Australia unlock the benefits of data and technology without compromising trust, safety or fairness?
Meta’s answer: not with the current approach.
In its submission, the tech giant argued that the Government’s proposed changes to the Privacy Act may go too far, particularly by limiting the ability to use publicly available user data to train AI systems.
“The development of effective generative AI models requires large and diverse datasets. Meta has found that the nature of AI technology necessitates the processing of comprehensive real data to avoid biases and ensure language accuracy,” Meta wrote.
“For example, synthetic data is not workable because it needs to be generated by a generative AI model itself; therefore, in order for a model to even create synthetic data, it needs to first be trained on an appropriate data source. Generative AI models need to understand common concepts such as language, visual elements, and abstract ideas.
“It is worth bearing in mind that, although there are databases of information that may not contain personal information, like Australian legislation, there are limits to the utility of such corpuses. Human beings’ discussions of culture, art, and emerging trends are not borne out in such legislative texts, and the discourse that takes place on Meta products both represents vital learning on both how individuals discuss Australian concepts, realities, and figures, as well as, in particular, how users of our products engage.
“This means that authentic and effective learning to ultimately power meaningful products of communication is best realised from training that includes those discussions and artefacts themselves”.
Meta urged the Government to retain an “principles-based” and “risk-based” approach to the privacy act, warning that rigid consent-based models may clash with innovation and content moderation laws.
Big businesses back innovation, but want clarity
While Meta’s submission has drawn attention for its direct challenge to privacy reform, many other large organisations echoed similar concerns, calling for national consistency, international alignment and less red tape.
Google cautioned against “overly broad or prescriptive” regulation that could deter AI adoption in sectors like health, agriculture, and logistics, while Amazon urged alignment with global data frameworks to avoid compliance burdens and unlock cloud infrastructure benefits.
Woolworths said data-driven tools like electronic shelf labels and digital trolleys had boosted productivity while supporting jobs. However, it warned that “uncertainty”, particularly inconsistent privacy and surveillance laws across states, was making it harder to innovate at scale.
“Privacy reform is also occurring at a time where surveillance device legislation varies greatly by State and Territory jurisdiction, and as a result, the protections for individuals and the obligations for organisations are inconsistent, often resulting in increased complexity and uncertainty,” the retailer wrote.
Woolworths backed the idea of a National AI Coordinator to align government policy and engage with international standards bodies like ISO.
Telstra warned that fragmented AI and privacy rules could stall digital investment and innovation, particularly in regional Australia, where telecommunications data could support breakthroughs in transport, disaster response and health services, if used safely.
As the Productivity Commission prepares to deliver its findings, one thing is clear: industry is calling for innovation-friendly policy, but not at the expense of trust, consistency or clarity. Whether regulators consider that kind of discourse fair and appropriate for AI training could define the next chapter of Australia’s data, privacy and AI landscape.
