From Personal Memories to AI Training Data: Navigating the New Frontier of Digital Legacy

To look at this article again, go to My Profile and then click on View saved stories.

I'm sorry,

My Memories Now Serve as Meta's Learning Material

In the narrative of "The Hopkins Manuscript" by R. C. Sherriff, the story unfolds in a universe 800 years post a devastating event that obliterated Western civilization. As researchers from a newly established society delve into the mystery of a forgotten era on their planet, they stumble upon the journal entries hidden in a marshy ruin once identified as England. For the citizens of this newly formed realm, the mundane details of a former school teacher's life in the countryside, along with his trivial conceits and efforts to cultivate award-winning poultry, become their window into understanding the culture of 20th-century Britain.

Once, I entertained the notion that if tasked with educating future entities about earthly existence, I could create a time capsule with far greater depth than the limited perspective of Sherriff’s character, Edgar Hopkins. However, upon reviewing my Facebook posts from ten years ago, I've come to consider that my own contribution might turn out to be even less inspiring.

In a recent statement, Meta revealed plans to preserve the adolescent musings and various posts of its users as a digital legacy for forthcoming AI generations. Starting June 26, a vast collection of historical public posts, vacation snapshots, and the identities of countless Facebook and Instagram users globally will be curated as a snapshot of human culture, repurposed into datasets for AI learning.

This implies that my everyday updates about college assignment deadlines (“Downed 3 energy drinks, 1,000 words to go”) and my typical vacation photos (including one where I'm dozing off with my phone on a motionless ferry) are set to become part of this collection. The boring and highly personal nature of these moments makes Meta's interest in them even more disconcerting. The company maintains that it's only focusing on content that is already public, meaning it won't touch private conversations, posts shared just with friends, or Instagram Stories. However, AI is now delving into personal moments that have long been forgotten in the far reaches of the internet. For those outside Europe, this process has already been completed. The cut-off date set by Meta was exclusively for Europeans. The contributions of American users on Facebook and Instagram have been aiding the training of Meta’s AI systems since 2023, as stated by company representative Matthew Pollard.

Meta isn't the only entity utilizing my digital footprint for AI development. Reece Rogers of WIRED recently noticed that Google's AI search capabilities were replicating his written work. However, pinpointing the exact bits of personal data contributing to the evolution of future AI conversational agents proved challenging. Tracking down contributions made to various platforms over the years can be difficult. For instance, early social media pioneer Myspace was bought by Time Inc. in 2016, which itself was purchased by Meredith Corporation two years later. Upon inquiring about my historical account with Meredith, they informed me that Myspace had been offloaded to an ad company, Viant Technology. Sadly, an attempt to contact Viant through an email address found on their site was met with a dead end, as the email bounced back with a notification that the address was “not found.”

Inquiring about my previous accounts with existing businesses turned out to be more direct. Tumblr, which is under the ownership of Automattic, the company that owns WordPress, informed me that unless I had chosen not to, my publicly shared adolescent posts would be distributed to "a limited circle of content and research collaborators, including entities working on AI model training," as announced in February. YahooMail, my email provider for many years, communicated that a selection of my old emails—now "de-identified" and "compiled"—is being "employed" by an internal AI system for tasks such as summarizing email content. LinkedIn, a Microsoft subsidiary, also mentioned that my public contributions are aiding in AI training, though it was noted that certain "private" aspects within those contributions were omitted. The specifics of what these private details entail were not clarified by a representative from the company.

Authored by Mark

Authored by Dmitri Alperovitch

Authored by Matt Kamen

Authored by Kim Zetter

Apart from LinkedIn, it appeared that other companies were more hesitant to use algorithms to analyze work-related messages or documents I produced. Slack, an office communication platform, refuted earlier claims that it trained AI with user messages. Similarly, Microsoft stated that the data generated in its range of office applications—Word, Excel, PowerPoint, Outlook (previously known as Hotmail), and Teams—was not utilized for enhancing base AI models. Google also confirmed that its professional tools, including both the free and premium versions of Gmail, were exempt, though it acknowledged that YouTube content might be used for AI training purposes.

Meta is certainly not in isolation. However, the way it rebrands content from Facebook and Instagram stands out because of the vast number of users who share deeply personal events on these platforms. Since I'm based in Europe, my posts on Facebook and Instagram are currently beyond the grasp of Meta's artificial intelligence. This situation could change, however. The company's revelation about its intentions to train its AI triggered a new dispute with European data protection authorities, leading Meta to momentarily halt its strategy of utilizing the posts of European users, Britons included, for model training purposes.

Meta expressed frustration over its AI's need for European data to grasp local languages and traditions, while the Austrian privacy advocacy group NOYB saw the halt as a provisional victory. NOYB, a constant irritant to Meta, has lodged complaints in 11 countries, claiming that Meta failed to offer Europeans an explicit option to avoid becoming data for AI training. (Meta refutes this, stating there was a form users could complete.) "Our issue isn't with their deployment of AI technology," NOYB's representative Mickey Manakas explained to WIRED. "It's that they must do so in accordance with the law."

The duration of Meta's suspension of its initiatives remains uncertain. Moreover, the prevailing approach among technology firms regarding the management of individual data is evident. The fragments and recollections we leave online are being converted into data for machine learning. If you believe there's still an opportunity to safeguard your digital legacy, you're mistaken. Your past is already being processed by the forces that will dominate our tomorrow.

Temporal Leap

A full nine years after receiving a life sentence, Silk Road founder Ross Ulbricht continues to be a symbol of significance within the libertarian circle in the United States. The modest slice of votes that the Libertarian Party managed to secure in the 2020 presidential race—amounting to 1.2 percent—was significant enough to draw the attention of Donald Trump, who has been actively engaging with its supporters. "Should you choose to support me, I promise that on my first day in office, I will alter Ulbricht's life sentence to time already served," Trump declared to an enthusiastic crowd at the Libertarian National Convention the previous month.

The possibility of Ulbricht being released motivated me to explore the WIRED archives to review Joshuah Bearman’s comprehensive account of Silk Road's emergence and collapse. Below is a snippet that highlights Ulbricht and his platform's foundation in libertarian principles.

Authored by Mark

By Dmitri Alperovitch

Authored by Matt Kamen

Authored by Kim Zetter

In pursuit of his goal, Ross Ulbricht experienced a moment of revelation. He penned in his diary, "The concept was to establish an online platform enabling individuals to make purchases anonymously, leaving no trace that could be traced back to them." He noted that he had been exploring the technology for some time but was in search of a viable business model and strategy.

Ross, aligning with libertarian views, saw drug consumption as a matter of individual freedom. He also noticed, as did many others, the utter ineffectiveness of the drug prohibition efforts. Thus, he decided that narcotics would be the ideal product for his budding venture. Initially naming it "Underground Brokers," Ross eventually opted for "Silk Road."

In many ways, Silk Road can be seen as a natural progression of the libertarian ideals that fuel a significant portion of the Internet (alongside the growing libertarian movement in U.S. politics). It represented the epitome of Silicon Valley's ethos, marrying disruptive technology with a political manifesto. Under the pseudonym Dread Pirate Roberts (DPR), Ulbricht positioned himself as the visionary leader of this movement, aiming to create a digital economy beyond the reach of state control, with Silk Road marking the first step toward an idealized libertarian utopia. Silk Road was not just a defiance of law enforcement; it was, as DPR asserted, a direct confrontation with the established power hierarchy itself.

This only added to the government's desire to close it. Ross felt honored by the unexpected media spotlight in June 2011, yet, he became worried when US Senator Charles Schumer held a press conference to criticize Silk Road. He expressed his concern, stating, "The US government, which I consider my primary adversary, knows who I am and is … demanding my downfall."

Inquire About Anything

Karl Emil queries, "Why would any politician think about supporting widespread surveillance?"

Karl Emil, thank you for raising this issue. You're referring to a proposed EU legislation aimed at scanning personal communications for signs of child sexual exploitation. This could involve the analysis of images on WhatsApp, text messages on iMessage, and videos on Snapchat. The discussion surrounding this legislation is particularly intense among a niche group in Europe, focusing on whether such oversight should extend to encrypted messaging services like WhatsApp and Signal. Karl Emil's inquiry seems to be posed with a hint of irony, given he uses the term "chat control 2.0" – a label given by privacy advocates who argue the legislation equates to widespread surveillance. Additionally, Karl Emil, with his background as an IT security expert, has voiced his opposition to the bill in the Swedish press.

Authored by Mark

Authored by Dmitri Alperovitch

Authored by Matt Kamen

Authored by Kim Zetter

Despite this, I'm compelled to address the query directly: What drives a lone lawmaker to support such a measure? Last year, I had the opportunity to speak with Ylva Johansson, the Swedish official responsible for internal affairs, who has fervently advocated for the legislation, to understand her perspective. Johansson takes great pride in the legislation, even referring to it affectionately as “my proposal.” She views herself as a warrior battling against large corporations that dodge regulations. “I believe it's my ethical duty to take action,” she explained to me. “Otherwise, what does that make me? Just a timid creature. Insignificant.” She argues that since people generally do not object to the use of spam filters that sift through emails for unwanted content, they shouldn’t oppose applying a similar approach to detecting child abuse material. However, privacy advocates and those behind encrypted messaging services counter that the technology doesn't operate in that manner. They argue that scanning would essentially create a loophole in what were previously secure communications. “To be perfectly clear,” states Andy Yen, founder and CEO of Proton. “Any widespread scanning effort weakens the integrity of encryption.”

On Thursday, it became apparent that disagreements persisted among European Union member states, leading to a conflict ahead of a planned vote on the proposal. Indications that Austria, Czech Republic, Germany, the Netherlands, and Poland were preparing to either vote against or not participate in the vote resulted in its delay.

Please send your inquiries to mail@wired.com and include "ASK LEVY" in the email subject.

Apocalyptic Times Bulletin

This week in Greece, amid consecutive heatwaves sending temperatures soaring past 100 degrees Fahrenheit, three visitors lost their lives.

In Conclusion

A recent publication raises the question: rather than extracting fresh minerals from the ocean floor, why not simply reuse the ones we already possess?

Should you find yourself seeking entertainment during the late hours at Denver’s airport, consider checking out Strava.

Authored by Mark

By Dmitri Alperovitch

Authored by Matt Kamen

Authored by Kim Zetter

When a duplicate of the streets encircling Taipei's governmental area materialized in the desolate area of Inner Mongolia, situated in northern China, it sparked curiosity about the real-life implications of a Chinese invasion of Taiwan.

Airbnb is anticipating that the Olympic Games will enhance its visibility in Paris and persuade the remaining doubtful residents.

Ensure you don't miss out on upcoming editions exclusive to subscribers of this column. Take advantage of a special 50% discount for Plaintext readers by subscribing to WIRED today.

Suggested for You …

Direct to your email: Dive into the future with Will Knight's Fast Forward, highlighting progress in artificial intelligence.

Inside the largest sting operation ever conducted by the FBI

The WIRED AI Elections Initiative: Monitoring over 60 worldwide electoral events

Ecuador finds itself completely without recourse amidst a severe drought.

Have peace of mind: These are the top-rated mattresses available for online purchase.

Luca Zorloni

Roger Reece

Dave Paresh

Steven Levy

Paresh Dave

Kyle MacNeill

N/A

Knight Will

Additional Content from WIRED

Evaluations and Tutorials

© 2024 Condé Nast. All rights reserved. Purchases made through our website may generate revenue for WIRED via our Affiliate Partnerships with retail stores. Reproduction, distribution, transmission, storage, or use of material from this site is strictly prohibited without prior written consent from Condé Nast. Ad Choices

Choose a global website