In May, TikTok announced a new feature that labels AI-generated content on its platform. However, this labeling system is not applied to all versions of the app. A recent report by the Mozilla Foundation and AI Forensics revealed that TikTok Lite-Save Data version, designed for users in less affluent markets, lacks key safeguards such as labeling AI-generated content and other safety measures.
According to Odanga Madung, a Mozilla fellow and coauthor of the report, labeling plays a crucial role in establishing trust and safety within online platforms. Unlike the full version of TikTok that provides labels for graphic or dangerous content, as well as information on topics like elections and health, TikTok Lite users do not benefit from these protective measures. This discrepancy means that users in poorer markets are exposed to deceptive content without clear indications of what is real or fake.
Madung questions why TikTok decided to prioritize cutting safety features in the Lite version of the app instead of optimizing other aspects. The company refuted the report’s claims, stating that content violating their rules is treated the same across both versions of the app and that various safety features are available.
Lite versions of apps are commonly introduced to cater to users in regions with limited data access or less advanced devices. TikTok Lite, launched in 2018 in Thailand and later expanded to Southeast Asian markets, has gained over 1 billion downloads. These Lite apps are crucial for companies to reach low-income users who are resource-constrained, as data accessibility plays a significant role in the AI-driven market.
Researchers discovered disparities between the full version of TikTok and TikTok Lite, particularly in how Covid-19 content is presented. While Lite users are directed to a resource center when searching for virus-related terms, actual Covid-19 content lacks proper labeling in the Lite version. Additionally, captions are shortened in TikTok Lite, reducing the context for users. Features like time limits and restricted mode, available in the full version, are absent in TikTok Lite.
By omitting these essential features, TikTok risks repeating the mistakes of other Lite apps like Facebook Lite. Biddle, a former fellow at Harvard’s Berkman Klein Center, notes that Facebook Lite’s removal of key elements hindered users’ ability to discern real from fake content. Madung expresses concern that the absence of labels and moderation in TikTok Lite could lead to the spread of misleading content beyond the platform.
In conclusion, the discrepancies between TikTok’s full version and TikTok Lite raise important questions about safety and user empowerment. As technology continues to evolve, it is crucial for companies to prioritize user safety and transparency across all versions of their platforms.