‘Next gold rush’: Australians’ data being used to train AI

An Australian expert has warned about how little Aussies know about their data being used to train AI.

Social media giants like Meta and LinkedIn have come under fire this month for using users’ data to train AI models, but one Australian expert warns it is happening a lot more than most people realise.

A LinkedIn setting called ‘Data for Generative AI Improvement’ has been automatically switched on for Australian users, giving permission for LinkedIn and unnamed “affiliates” to “use your personal data and content you create” on the social network “for training content creation AI models”.

Meanwhile, Meta has admitted to scraping Australian adult users’ public data – like photos and posts on Facebook and Instagram dating back to 2007 – to train its generative AI models.

Dr James Birt, an associate professor of creative media at Bond University, who did his PhD in AI and software engineering, says Australians need to understand that every platform they are on uses their data.

Meta has admitted to scraping adult users’ public data on Facebook and Instagram to train its AI models. Picture: Chris Delmas / AFP

“That is the truth; they are using your data to train these systems,” he told news.com.au.

He added: “There’s no such thing as a free lunch, so you have to be aware that nothing is free. What you are giving up is essentially your data to be able to be used in various different ways.”

He explained when using online platforms, people who care about their data security should always explore a way to opt-out.

“You’re always opted in, you yourself need to see if there is a way to opt out,” he said.

“It might be an easy way, it might be a convoluted way, there may not be a way, but it is something one needs to educate themselves about.”

Dr Birt urged people to read the fine print on everything from social media to online children’s games, and even medical consent forms following reports last week that medical scans from an Australian radiology company were being used to train AI.

“If you’re going to the doctor or getting X-rays, what is actually being sent off?” he urged people to consider.

“Is there a way to tick the box to say ‘I do not wish for my data to be used in these training models’ or some people may say ‘I’m happy for my data to be used because it might assist in patient care in the future.’”

A LinkedIn spokeswoman told news.com.au that their opt-out setting was because they believe ‘members should have the ability to exercise control over their data’.

He described AI training data sets as “the next gold rush”.

“Predominantly, these models have been trained on textual data, now it’s moving into imagery, video and audio, and then there’s a whole future related to unconscious data that perhaps we’re not even aware of what we’re providing,” Dr Birt said.

“That’s that next generation of using VR headsets and sensor technology where we’ve got heart rate, brain wave and pulse patterns … we really don’t know how these data sets are perhaps being used or will be used, but they are being gathered, and they are being stored, and they are being used to train models that we might see in the next couple of years.”

He said a lot of companies were collecting data without yet knowing how it will be used.

He said education and awareness was crucial in today’s rapidly changing tech landscape because while there are now classes covering these topics at school, most Australians have never learnt how to protect their data online.

“It moves so quickly and changes almost on a daily basis that is very difficult for people to keep up with the ebbs and flows,” Dr Birt said.

He said while regulation was incredibly important to protect Australians, the reality is that tech users need to do what they can to protect themselves because the law “trails the tech industry” and is constantly made “in hindsight”.

“The tech industry innovates and drives it, and the law always drags behind,” he said.

UN experts’ AI warn warning

United Nations experts last week cautioned that the development of artificial intelligence should not be guided by market forces alone.

They held back from suggesting the formation of a muscular worldwide governing body to oversee the rollout and evolution of the technology.

The panel of around 40 experts from the fields of technology, law and data protection was established by UN Secretary-General Antonio Guterres in October.

Their report raises alarm over the lack of global governance of AI as well as the exclusion of developing countries from debates surrounding the technology.

“There is, today, a global governance deficit with respect to AI,” which by its nature is cross-border, the experts warned in their report.

UN Secretary-General Antonio Guterres. Picture: Angela Weiss / AFP

“AI must serve humanity equitably and safely,” Mr Guterres said.

“Left unchecked, the dangers posed by artificial intelligence could have serious implications for democracy, peace and stability.”

Against the backdrop of his clarion call, the experts called on UN members to put in place mechanisms to ease global co-operation on the issue, as well as to prevent unintended proliferation.

“The development, deployment and use of such a technology cannot be left to the whims of markets alone,” the report said.

US accuses social media giants of ‘vast surveillance’

Also last week, the US Federal Trade Commission revealed that a years-long study showed social media titans have engaged in “vast surveillance” to make money from people’s personal information.

A report based on queries launched nearly four years ago aimed at nine companies found they collected troves of data, sometimes through data brokers, and could indefinitely retain the information collected about users and non-users of their platforms.

“The report lays out how social media and video streaming companies harvest an enormous amount of Americans’ personal data and monetise it to the tune of billions of dollars a year,” FTC chair Lina Khan said.

“Several firms’ failure to adequately protect kids and teens online is especially troubling.”

Ms Khan contended that the surveillance practices endangered people’s privacy and exposed them to the potential of identity theft or stalking.

The findings were based on answers to orders sent in late 2020 to companies including Meta, YouTube, Snap, Twitch-owner Amazon, TikTok parent company ByteDance, and X, formerly known as Twitter.

– with AFP