The social media platform X may be breaching Australian privacy law.

Recent changes automatically opt users into having their posts used to train an AI system, Grok AI. 

Australia's privacy watchdog, the Office of the Australian Information Commissioner (OAIC), has raised concerns, but stopped short of launching a formal inquiry.

“We are increasingly seeing digital platforms using the personal information they collect for one purpose for the secondary purpose of training generative AI models. This gives us cause for concern about whether practices comply with the Privacy Act,” the OAIC says

X users have been concerned by a pre-ticked box in the app's privacy settings that allows X to use posts for AI training. 

Under Australian law, platforms must ensure default settings enable user control and seek explicit consent for data use.

John Pane, chair of Electronic Frontiers Australia, says X's practices “appeared to breach Australian privacy law”.

However, he also noted the challenge OAIC faces in enforcing Australian law against X, a US-based company, due to weaker US federal data privacy laws.

Other platforms like Meta and Slack are also harvesting user data to train AI, part of a global race to develop large language models (LLMs). 

Pane pointed out that platforms are “feverishly rewriting privacy policies” to include data use for AI training.

More details are accessible here.