NurPhoto
·
gettyimages.comMeta's AI Data Grab
Curated by
cdteliot
3 min read
180
According to recent reports from Reuters, Meta is set to use public posts, including photos and comments, from adult users on Facebook and Instagram to train its generative AI models, raising concerns about privacy and data usage.
The Predicament of AI Training
Meta entered the AI arena as early as 2013, but back then, there weren't advanced AI models yet that could train themselves on human content. However, now it is not just a possibility but a surprising reality. Of course, Meta couldn't afford not to jump on the opportunity. They have both Facebook and Instagram, two of the biggest and most active social media platforms on the Internet. And what do these two have in common? Yes, that's right: human content.
1
It has long been unconfirmed but speculated that Meta is scraping its users' posts to train their AI models, such as Llama and Vizseq. Recently, though, we have a definite answer: Yes, Meta is indeed using everyone's posts to train their AI.2
1
2 sources
Meta’s Shocking (But Still Expected) Revelation
The answer came from an inquiry, where Australian senators cornered Melinda Claybough, Meta's global privacy director.
1
2
Tony Sheldon, the labor senator, first brought up the topic when he asked Claybough if the tech giant has been making use of his countrymen's posts for constructing Meta's AI libraries since 2007. Claybough denied the accusation, stating, "We have not done that."1
However, another senator, David Shoebridge, didn't let the subject slide. He confronted Claybough with a sharper question, "The truth of the matter is that unless you have consciously set those posts to private since 2007, Meta has just decided that you 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?"1
2
Claybough could not deny that one and simply replied, "Correct."1
2
2 sources
Why This is a Cause for Public Concern
This fact raises fears and worries about privacy, ownership, and consent since most people just post on either Facebook or Instagram without thinking twice about artificial intelligence lurking around the corner, waiting to get its wiry hands on human content. Most people aren't even informed that this is happening. Meta wouldn't explicitly ask for your permission every time they grab your content. Sure, it may be in the fine print of those long Terms and Conditions you didn't bother to read, but because you just scrolled through that lengthy list, you just unknowingly allowed them access and use of your posts by immediately ticking the "I agree" box.
1
2
You'll never know if that post you made yesterday and the posts you published on the days before that have already been cataloged and shoved into Meta's AI shelves. You may use their AI tools and see your face staring back at you from generated photos.3
4
4 sources
The Exceptions
Meta's AI training isn't absolute and a few fortunate countries have been spared from this turmoil. Of course, that's only possible because of government intervention. For European countries, Meta has put their AI operations on hold because of certain privacy laws that even a multi-billion dollar company can't go against.
1
If they push through, they would have to allow European users to opt-out of their AI system to comply with those privacy regulations.2
Brazil is another safe space from this AI surveillance as the nation thwarted any attempts of Meta to use the data of Brazilian citizens. Their government is firm: Meta can never claim content from Brazilian users to train their AI and there's no way around it.3
3 sources
The Extent of Meta’s AI Training
Now we know that Meta uses our content, both text and images, for their AI— but how far can they go? According to Claybough, they don't scrape the posts of minors aged under 18, but if you're an adult and you post your children on your own account, Meta would consider it up for grabs. If you're not careful, even your kids will be part of Meta's AI training.
1
2
Thankfully, there's a way out. Meta's AI only pulls public content. This means that if you set your posts to private or to a select audience, the system won't touch them and will leave them alone. However, if your posts are already public, it doesn't matter if you change their privacy settings now. Meta may have already processed them into its AI brain. Once your data is out there, there's no taking it back.3
4
4 sources
Closing Thoughts on Meta’s AI Training
Meta has leveraged its social media channels to gain access to user content, claiming everything that's free game for building their AI models. Unfortunately, there's no stopping them. Unless you're underaged or you're from Europe or Brazil, Meta can add your posts to their AI database.
1
2
The only solution is to change your privacy settings and minimize the reach of your posts. However, once Meta already claimed your content, even switching your post to "Private" won't erase it.3
Let this serve as a warning for those who don't want AI to meddle with their data. Meta already has its claws on your Facebook and Instagram accounts. You have to be vigilant of what you publish on these platforms, and only share what you think is appropriate. Before you click that "Post" button, always be aware that AI is just waiting to snatch your content from under your nose and use it for its own gain.4
4 sources
Related
How can I opt out of Meta using my data for AI training
What specific data does Meta collect from users for AI training
How does Meta ensure the privacy of users' data in its AI models
What are the consequences of not changing privacy settings on Meta platforms
How does Meta's AI training affect user trust in its services
Keep Reading
Meta's AI Strategy: A Comprehensive Overview
Meta, the parent company of Facebook, Instagram, and WhatsApp, is making significant investments in artificial intelligence to power its products and services. With a focus on machine learning, computer vision, natural language processing, and other key AI technologies, Meta aims to enhance user experiences, improve content moderation, and unlock new possibilities across its family of apps.
17,156
Meta Works on Paid AI Assistant
Meta Platforms is exploring the possibility of launching a paid version of its AI-powered assistant, Meta AI, as it aims to compete with other tech giants in the rapidly evolving AI market. This move comes as Meta continues to invest heavily in AI and seeks to establish itself as a leader in the field.
36,951
The Rise of AI-Generated Pornography
The rise of AI-generated pornography has sparked intense debate over its ethical implications, technological advancements, and societal impact. This emerging phenomenon leverages sophisticated deep learning algorithms and generative models to create highly realistic adult content, raising concerns about consent, exploitation, and the blurring of reality and fantasy. As AI technology rapidly evolves, it presents both potential benefits, such as reducing human exploitation in the porn industry,...
6,996
Zuckerberg: AI Data Overvalued
Meta CEO Mark Zuckerberg has sparked controversy by suggesting that creators and publishers overvalue their contributions to AI training data, a stance that aligns with the broader tech industry's perspective on the use of vast amounts of information for developing artificial intelligence systems.
7,581
