San Francisco, California – TikTok made waves in the social media world by becoming the first platform to automatically label artificial intelligence-generated content. This move comes in light of increasing concerns surrounding the spread of online disinformation and deepfakes, as AI technology continues to advance at a rapid pace.
Online giants such as Meta, the parent company of Facebook, and TikTok had already implemented rules requiring users to disclose if their content was created using AI software. However, TikTok took a significant step forward by announcing that it will now label videos identified as AI-generated. This labeling will apply to content produced using various AI tools, including Adobe’s Firefly tool, TikTok’s own AI image generators, and OpenAI’s Dall-E.
According to Adam Presser, TikTok’s head of operations and trust and safety, the platform’s decision to label AI-generated content is a response to the growing prevalence of harmful AI-generated content. He emphasized the importance of authenticity within TikTok’s community, highlighting the desire for users to distinguish between human-created content and AI-enhanced or generated material.
As social media platforms like TikTok and Meta delve deeper into integrating generative AI into their services, concerns have arisen over the influx of low-quality AI-generated spam content flooding users’ feeds. With major elections taking place worldwide, these companies face mounting pressure to implement safeguards against misleading deepfakes, covert influence operations, and improper content moderation while maintaining neutrality.
In a bid to address these concerns, TikTok and its parent company, ByteDance, recently filed a lawsuit against the US government challenging legislation aimed at either forcing the sale or banning the app. Lawmakers had expressed worries about the potential for TikTok to disseminate disinformation and propaganda. TikTok also announced its partnership with a coalition led by Adobe to incorporate content credentials into AI-generated products, aiming to embed digital fingerprints into multimedia AI content to identify its source, timing, and creators.
OpenAI, a prominent player in the AI industry, revealed its plans to embed fingerprinting technology into all images generated by its models, including the widely anticipated video-generating model. This move signifies a step toward greater transparency and accountability in the era of AI-generated content.
Notably, other tech giants like Google, Microsoft, and Sony are also exploring ways to integrate fingerprinting technology into their AI tools. Meta, for instance, announced its intention to label AI-generated content with a “Made by AI label” by detecting invisible markers inserted by various AI groups like Google, OpenAI, and Microsoft. However, experts caution that bad actors may still utilize open-source AI tools to create undetectable deepfakes, underscoring the ongoing challenges in combatting digital manipulation.
Overall, TikTok’s initiative and the wider industry push towards transparent AI-generated content represent crucial first steps in addressing the complexities of disinformation and deepfakes in the digital age. Dana Rao, general counsel and chief trust officer at Adobe, highlighted the significance of transparency in fostering authentic digital conversations amidst a landscape where digital manipulation is increasingly prevalent.