It’s more challenging to tidy up a mess while you’re still contributing to it.
As 2025 concluded, Instagram head Adam Mosseri expressed concern about AI’s impact on reality. “Authenticity is now easily replicated,” Mosseri noted. “The unique qualities of creators—genuineness and a unique voice—are now accessible to anyone with the right tools.” However, Mosseri believes people still seek “content that feels real” and suggested labeling real media with cryptographic signatures from camera manufacturers to ensure authenticity.
The solution Mosseri proposed already exists: C2PA. Nevertheless, Instagram’s use of it seems ineffective, as they continue developing generative AI tools. AI’s ability to mimic reality threatens social media platforms built on creators’ content. AI easily replicates trending styles, making raw aesthetics less distinguishable. More worryingly, AI can propagate misinformation about significant events.
Several major tech companies have endorsed C2PA (Content Credentials) to verify non-AI-generated media through invisible metadata. Meta joined C2PA in 2024, emphasizing the importance of understanding digital content. Despite support from companies like Microsoft, Meta, and Google, C2PA struggles with practical implementation. People remain largely unaware of the system, which does too little to combat AI-generated misinformation.
Companies have also adopted tools like Google’s SynthID, but provenance-based solutions face obstacles, including universal adoption challenges. Leica Camera USA emphasized that older cameras still rely on context and reputation for trust. Provenance metadata is prone to accidental or intentional removal, making it unreliable. Platforms like YouTube and TikTok struggle with inconsistent tagging of AI-generated content.
Despite these challenges, C2PA continues to expand, albeit slowly. AI companies do not fully address the root issues while benefiting financially from AI tools. Efforts to verify content authenticity remain largely reliant on the honor system, while platforms like Instagram consider focusing on who says something rather than the content itself. In essence, AI’s challenge to authenticity persists in a landscape where solutions are underway but far from comprehensive.