TikTok Introduces AI Content Limiter for 'For You' Feed

TikTok has become the latest social media platform to empower users with greater control over their content experience, introducing a new option to limit AI-generated content in their 'For You' feeds. The platform is rolling out a dedicated slider within its 'Manage Topics' tool, allowing users to actively reduce the presence of AI-created videos and images.

This new AI-generated content slider integrates into TikTok's existing 'Manage Topics' control panel, a feature first launched in June. 'Manage Topics' provides users with a mechanism to inform the TikTok algorithm about their content preferences, thereby refining their personalized feed. The addition of this slider specifically targets AI-generated clips, offering a direct way to manage their visibility.

According to TikTok, the new setting functions similarly to existing controls within 'Manage Topics,' which allow users to adjust the frequency of content from over ten categories like Dance, Sports, and Food & Drinks. TikTok states:

“Manage Topics already enables people to adjust how often they see content related to over 10 categories like Dance, Sports, and Food & Drinks. Like those controls, the [AI content] setting is intended to help people tailor the diverse range of content in their feed, rather than removing or replacing content in feeds entirely.”

This emphasizes that the feature aims to tailor the feed, not completely eliminate AI content.

Consequently, users who are weary of encountering uncanny AI-generated videos, such as those with unnatural acting or digitally altered performances, can now signal their disinterest to TikTok. While this should significantly reduce the appearance of such content in their 'For You' feed, it's important to note that some AI-generated clips may still appear. This is due to the current limitations of TikTok's detection systems, which cannot yet identify every piece of generative AI content.

To enhance its detection capabilities, TikTok is actively developing and testing "invisible watermarking" for generative AI videos. This technology aims to embed metadata within clips, indicating when AI tools have been used in their creation. TikTok announced:

“Over the coming weeks, we'll start adding invisible watermarks to AI-generated content made with TikTok tools like AI Editor Pro, and content uploaded with C2PA Content Credentials. The watermarks will help us label content more reliably, and give us more context around changes made to content.”

Currently, TikTok mandates users to label realistic AI-generated content and employs C2PA Content Credentials for in-stream identification. However, these existing measures face limitations, particularly when videos are cross-posted across platforms, as crucial metadata and markers are often lost during the transfer.

The platform hopes that a broader adoption of invisible watermarking, in conjunction with C2PA standards, will lead to more accurate tagging of AI content. This, in turn, is expected to significantly improve TikTok's ability to effectively limit AI-generated content for users who choose to opt out.

This move by TikTok aligns with a broader trend among social platforms. Pinterest recently introduced similar AI content restrictions, allowing users to opt out of what some describe as 'AI slop' in their feeds. These developments suggest that user enthusiasm for generative AI content may not be matching the high demand anticipated by many developers.

This trend stands in contrast to forecasts from companies like Meta, which have repeatedly envisioned a future where social feeds are increasingly dominated by AI-generated content. Meta has suggested that even social interactions could become more populated by AI bots, believing users will eventually prefer AI updates over human interaction.

While significant investments are being made in AI development, current user data suggests a more tempered interest in artificial visuals and video, despite their increasing realism. The fundamental nature of social media, designed for human connection and updates from known individuals, poses a challenge to the widespread adoption of AI content. It remains to be seen if future generations, growing up with ubiquitous generative AI, will develop different preferences.

The introduction of AI content controls by platforms themselves underscores this current user sentiment. While this could represent early resistance that may shift with future generations, there is an undeniable immediate demand for such controls. Other platforms are likely monitoring the popularity of these features, potentially leading to similar implementations. This trend serves as a significant indicator of both the adoption of AI and the existing resistance to generative AI elements within social media feeds.

Beyond content controls, TikTok is also investing in AI literacy and safety. The company is launching a $2 million fund to commission experts to create educational content on TikTok about AI. Furthermore, TikTok is sponsoring the Partnership on AI to support its initiatives in responsible AI development. The long-term impact of these user controls and educational efforts remains to be seen, with future data on user adoption providing valuable insights into evolving preferences for AI content.