Platforms' AI policy updates in 2024: labelling as the silver bullet?

Using a self-developed assessment framework, EU DisinfoLab tracked how Facebook, Instagram, YouTube, X and TikTok updated their policies on AI-manipulated and AI-generated content through 2024. All five platforms had signed a February 2024 voluntary pledge on election-related deepfakes and most participated in the C2PA content-provenance standard, but individual actions diverged: YouTube introduced a creator-disclosure tool relying on an honor system, Meta shifted toward labelling and contextualizing rather than removing content, TikTok began auto-labelling AI content sourced from other platforms, and X made no substantive policy changes.

The report argues that labelling alone cannot address the risks posed by AI-generated content and should be paired with other moderation measures, that platforms largely ignore AI-generated text despite its documented use in influence operations, and that enforcement criteria — such as YouTube's tagging rules — remain vague.

It concludes that responsibility is disproportionately placed on users and industry self-reporting rather than platform-side detection, and that the 2024 updates amount to only minor changes that do not meaningfully advance AI content governance.