Science and Nature news
Facebook parent company Meta has admitted its millions of Australian users can’t opt out of having their data being used to train the company’s artificial intelligence models, even as users in the European Union are able to do so.
Since changing its privacy policy for Australian users in June, Facebook has been hoovering up nearly two decades’ worth of public Facebook posts, including text, photos and status updates, to train its language model, dubbed Llama.
Appearing on Wednesday before a Senate inquiry into the adoption of artificial intelligence technologies, Meta’s global privacy director Melinda Claybaugh confirmed that Australian users do not currently have the option to opt out of the mass data collection.
Meta’s privacy policy director Melinda Claybaugh in 2023.Credit: Oscar Coleman
Users in the European Union, on the other hand, can refuse, thanks to that region’s stricter privacy laws.
“I want that same option as the EU, and I want to know why I can’t have it, and why Australians can’t have that same option,” Labor senator Tony Sheldon asked the Meta executive. “Will you give the Australian public the EU option of opting out?”
“The specific option that we’re offering in Europe [sic] was in response to a very specific legal framework,” Claybaugh said, adding that she could not speak to what might happen in the future.
“Is it happening now?” Sheldon asked, referring to whether Australians can currently opt out of the data scraping.
“It is not,” Claybaugh said.
Greens senator David Shoebridge also probed Meta executives over the company’s mass data scraping, noting Meta had chosen to train its AI model using text and images dating as far back as 2007.
“The truth of the matter is that unless you have consciously set those posts to private since 2007, Meta has just decided that [it] will scrape all of the photos and all of the texts from every public post on Instagram or Facebook since 2007, unless there was a conscious decision to set them on private. That’s the reality, isn’t it?,” Shoebridge asked.
“Correct,” Claybaugh responded.
“Meta must think we’re mugs”: Labor senator Tony Sheldon.Credit: Trevor Collens
Australia is weighing how best to regulate AI technologies, and Sheldon told this masthead after the hearing that the nation’s privacy laws may need an overhaul in response to the revelations. The federal government has flagged an update to the Privacy Act, but those changes have been delayed multiple times and are yet to be introduced to the parliament.
“Meta must think we’re mugs if they expect us to believe someone uploading a family photo to Facebook in 2007 consented to it being used 17 years later to train AI technology that didn’t even exist at the time,” Sheldon said.
“The personal moments people share – photos, videos, and records of people’s lives, their children, and their families – are not fodder for a tech company to turn into new products. It’s an unprecedented violation of what makes us human.
“It’s not just dishonest, it’s predatory. And if our privacy laws allow this, they need to be changed.”
‘It’s not just dishonest, it’s predatory. And if our privacy laws allow this, they need to be changed.’
Labor senator Tony Sheldon
Independent senator David Pocock also told this masthead that new privacy laws are needed.
“A clear takeaway from Meta’s testimony this morning is that we urgently need strong privacy legislation, like the European Union, to force these social media giants to do the right thing by our communities in Australia,” he said.
“I’m also really concerned about the total lack of transparency around social media algorithms and the harm they are doing, especially to young people.”
Senator David Pocock.Credit: The Sydney Morning Herald
Meta has been hit with complaints in 11 countries over its data collection practices, but says it’s simply following the same approach as other AI firms such as Google and OpenAI.
A Meta spokeswoman previously told this masthead that the company was “committed to building AI responsibly”.
“With the release of our AI experiences, we’ve shared details about the kinds of information we use to build and improve AI experiences – which includes public posts from Instagram and Facebook – consistent with our privacy policy and terms of service,” the spokeswoman said.
“While we don’t currently have an opt-out feature, we’ve built in-platform tools that allow people to delete their personal information from chats with Meta AI across our apps. Depending on where people live, they can also object to the use of their personal information … to build and train AI consistent with local privacy laws.”
Meta also defended a response by its chatbot that described Hamas as a Muslim “armed resistance” movement, but did not mention its October 7 massacres in Israel.
Loading
Meta’s regional public policy boss, Simon Milner, said he had not seen the response that opposition senator James McGrath read out in parliament, but viewed it as a “factual response”.
“You may have gone on to ask it more questions about the atrocities of October 7 last year and I would be interested to see how that played out,” Milner told the parliamentary inquiry.
Milner said Meta’s AI tool was learning and improving over time.
“The AI learns from feedback when people feel that there are issues around political bias, and we’re always looking to improve it,” Milner said.
The Senate committee into AI adoption is set to report to parliament before September 19.
with Nick Bonyhady
The Business Briefing newsletter delivers major stories, exclusive coverage and expert opinion. Sign up to get it every weekday morning.
Most Viewed in Technology
Loading
