A Deep Dive into AI Labels on Social Media

As agencies begin to leverage AI tools in creative production, it’s important to know what limitations online platforms are introducing to monitor its use. Social media platforms have begun adapting their policies and steadily introducing labels that indicate the use of AI-generated content. This is in an effort to improve transparency and combat misinformation online.

Here’s a quick overview of how different platforms are addressing the use of AI, based on recent insights shared within our team. This guide aims to provide a comprehensive resource for FUSE Create and other creators and advertisers navigating these changes. Plus, there will be a cheat sheet at the end for all the marketers out there!

General Overview

Using in-platform AI tools on social channels often results in automatic content flagging. However, the approach varies for third-party tools. Meta is currently the most stringent, requiring labels to avoid penalties. LinkedIn and TikTok, along with Microsoft, Google, Adobe, and OpenAI, have adopted the Coalition for Content Provenance and Authenticity process (C2PA), a standard that outlines the specifications for establishing content provenance and authenticity of content.

Platform-Specific Guidelines

TikTok

AI-Generated Labels: TikTok will automatically apply an AI-generated label if it detects third-party tools. Creators can also disclose AI use themselves.

Disclosure: Creators can label content as AI-generated by adding text, a hashtag sticker, or context in the video’s description. Or by applying the label in content settings.

Label Requirements:
– Mandatory for content with realistic images, audio, and video.
– Encouraged for fully AI-generated content or significantly AI-edited content.

Meta

AI-Generated Labels: Meta may add an “Imagined with AI” watermark if its AI tools are used, even if the content is re-uploaded. Third-party AI use may trigger an AI info label.

Disclosure: Content must be labeled in the post/story/Reel settings before sharing.

Label Requirements:

– Required for photorealistic video or realistic-sounding audio that is digitally generated or altered.

– Penalties may include content removal, especially for paid media.

LinkedIn

AI-Generated Labels: Tags (small C2PA logo) are automatically added and identified by the C2PA process.

Disclosure: Currently, there is no way for creators to label content themselves.

Label Requirements: Limited information on specific requirements.

YouTube

AI-Generated Labels: Creators must add the AI label during the upload process. YouTube may also apply the label if not disclosed.

Disclosure: Part of the upload process.

Label Requirements:

– Required for content that is meaningfully altered or synthetically generated and appears realistic.

– Penalties for repeated offenses, with higher risks for paid media.

Other Platforms

With Snapchat, they label content that is shared using in-platform AI and/or filters that use AI. Pinterest has no specific AI-labeling requirements, similar to Reddit. However, there are subreddit-specific rules that are enforced by moderators (so make sure to read the rules carefully!).

Recommendations

For Meta and YouTube, it’s advisable to label content as AI-generated, even if not technically required. Meta is likely to tag paid content automatically, and YouTube’s strict enforcement and challenging appeal process make proactive labeling a safer choice. Especially for retailers; the marketing landscape is undergoing a monumental shift due to the integration of AI in consumers online shopping journeys. Making sure you are following these labeling rules will help you stay compliant.

By staying informed with these evolving guidelines, creators can navigate the complexities of AI-generated content and maintain their presence across various platforms. For your own records, we created handy cheat sheet for all the social media marketers on your team 😉