TikTok Lite’s Safety Features Fall Short, Leaving 1 Billion Users at Risk

To go back to this article, head to My Profile and then click on View saved stories.

TikTok Lite Compromises Safety Features for Nearly 1 Billion Users

In a May announcement, TikTok disclosed plans to mark content produced by artificial intelligence on its site. However, this policy does not apply across all iterations of the application. Recent findings by the Mozilla Foundation and AI Forensics reveal that TikTok’s Data-Saving Lite version, targeting users in less affluent regions, fails to identify AI-created content and is missing additional, comparable security measures.

"Platforms utilize labeling as a crucial strategy to provide a sense of trust and security," states Odanga Madung, a Mozilla fellow and report coauthor.

Individuals utilizing the complete edition of TikTok, as an instance, will encounter warnings highlighting the presence of explicit content or actions that could be hazardous. Likewise, materials discussing subjects such as political elections and wellness issues will feature an advisory prompting users to visit a "resource hub" within the application for reliable information.

On TikTok Lite, these protective measures are absent. This includes, among various concerns, the problem of AI-generated content misleading users during elections worldwide. Consequently, individuals in less affluent regions receive less guidance on distinguishing between fake and authentic content compared to those in wealthier areas.

Madung is puzzled as to why, out of all possible features to eliminate for app optimization, the company decided to reduce those that enhance user safety on the platform. "It's unclear whether this was an intentional decision or simply oversight," he remarks.

"A spokesperson for TikTok highlighted multiple errors in the report, stating that it incorrectly portrays the company's commitment to safety. They emphasized, "In reality, any content violating our guidelines is equally removed from both TikTok Lite and our primary application, and we provide a wide array of safety measures." However, the company did not specify the exact mistakes in question.

Slimmed-down versions of applications have been a strategy for businesses to expand their reach in regions where high data charges prevail or consumers can only afford more basic smartphones. In 2015, Meta, previously known as Facebook, rolled out Facebook Lite, a simplified variant of its application better suited for 2G networks. In the same year, it introduced Free Basics, a program enabling users in the Global South to use the platform and some other websites without incurring data costs. (Such apps or services are referred to as "zero-rated.") At launch, the initiative was widely criticized, especially in India, for offering an inferior experience to less affluent users.

In 2018, TikTok introduced a Lite variant of its app starting in Thailand, quickly spreading its reach to additional Southeast Asian countries such as Indonesia, Malaysia, and the Philippines. This version is designed to operate efficiently on 2G and 3G networks, unlike its full-featured counterpart. As per information from the Google Play Store, TikTok Lite, which is exclusively available for Android devices, has surpassed 1 billion downloads.

"Most users in the Global South have limited income and resources," observes Payal Arora, a professor specializing in inclusive AI cultures at Utrecht University. She notes that simplified app versions are key for engaging these individuals, a strategy that has become increasingly important. Arora explains, "In today's market, driven and fueled by AI, data acts as a form of currency."

Researchers identified various distinctions between the app versions, notably concerning Covid-19 information. Users of the Lite version, when searching for virus-related terms, are directed to a resource center. However, unlike in the full TikTok version, this content isn't accompanied by a link to the resource center.

Investigators discovered that in TikTok's Lite version, captions were cut short, providing viewers with limited information about the content. Additionally, this scaled-down version of TikTok does not include a tool for users to control their usage duration or a "restricted mode" designed to filter out content potentially unsuitable for all viewers, as per TikTok's guidelines.

By removing these aspects, TikTok seems to be mirroring the errors made by earlier Lite versions of apps. For example, Facebook Lite also eliminated components that could have aided users in more effectively comprehending the content they encountered.

Ellery Roberts Biddle, who has researched zero-rated social networks during her fellowship at Harvard's Berkman Klein Center for Internet and Society, pointed out that the Lite version of Facebook presented challenges in distinguishing between authentic and false information. This was mainly because it frequently omitted images due to their large size or restricted the ability to access links. Biddle believes these issues stem from an apparent disregard for empowering the users.

Madung expresses concern that the issues stemming from insufficient labeling and oversight could extend beyond just TikTok Lite users. He points out, "The creations on TikTok often find their way beyond the app," noting that it's common for users to save videos and circulate them on other platforms like WhatsApp or Instagram Reels.

In a related development …

Insider Information: An email was sent to Joe Biden's campaign staff indicating his withdrawal from the race, instructing them to consult platform X for details.

J.D. Vance's Venmo account was open to the public, revealing its contents.

In your email: Receive Plaintext—Steven Levy's in-depth analysis on the worldwide influence of technology

Apple, Nvidia, and Anthropic have harnessed thousands of videos taken from YouTube to educate artificial intelligence systems.

Surprisingly, purchasing generic versions of Ozempic on the internet is quite straightforward, regardless of whether you actually require it or not.

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

Additional Content from WIRED

Insights and Tutorials

© 2024 Condé Nast. All rights reserved. Purchases made via our website may result in WIRED receiving a share of the sale through our affiliate agreements with retail partners. Content on this website is protected and cannot be copied, shared, or used in any form without explicit consent from Condé Nast. Ad Choices

Choose a global website