Кризис доверия: почему ярлыки на ИИ-контенте не спасут реальность

Remember when video footage from the scene was irrefutable proof? Forget it. We have officially entered an era where skepticism becomes the default setting of perception. Instagram's head Adam Mosseri recently dropped an information bomb, stating that society needs to stop treating photographs and videos as accurate snapshots of reality. This isn't just a private opinion—it's an admission of capitulation by major platforms in the face of a deepfake wave.

At the heart of this battle for truth stands the C2PA standard, or Content Credentials. Adobe led this initiative, bringing Microsoft, Google, and even OpenAI as allies. The idea seemed sound: embed a "digital passport" in every file that would reveal when a photo was taken, who edited it, and whether artificial intelligence had a hand in it. But in practice, the system resembles a leaky sieve. OpenAI itself admits that C2PA metadata is incredibly easy to remove—intentionally or accidentally when uploading to social networks.

The problem is that the industry has never managed to agree on the rules of the game. While Google implements these labels in its Pixel smartphones, Apple—the world's leading camera manufacturer—maintains mysterious silence. Without iPhone's participation, any attempt to create a unified authenticity standard is doomed to fail. Moreover, nobody is rushing to update old cameras. Photographers aren't running to buy new bodies just to get a "verified" checkmark, and manufacturers like Sony and Nikon are extremely reluctant to share their plans for firmware updates on older models.

The situation is compounded by the blurriness of the very concept of "photograph." We've been living for five years now in a world of computational photography, where pressing the shutter button launches dozens of processing algorithms. Night mode, frame stitching, HDR—is this already AI or not yet? Google now claims that the goal of a camera is to capture "memories," not moments in time. When the boundaries of reality are so blurred at the hardware level, demanding clear "truth/falsehood" labeling from software becomes meaningless.

The creative community isn't thrilled either. The label "Made with AI" is often perceived as a stigma that devalues the artist's work. Platforms like Instagram have already tried slapping such badges on everything, but quickly retreated under a barrage of criticism. Content creators are furious, users are confused, and social media algorithms continue to breed "information garbage" because it brings views and profit. Corporations like Meta and Google have a conflict of interest: they spend billions developing AI while simultaneously needing to convince us that this very AI is a threat that needs to be labeled.

We find ourselves at a strange juncture where even governments use AI manipulation for propaganda, while tech giants throw up their hands. The hope that metadata will save our collective understanding of reality is fading. We will likely face strict regulation from governments, but even that won't return to us the simplicity with which we once trusted the frames in our news feeds. Reality isn't just changing—it's becoming a matter of faith, not facts.

The key point: The C2PA standard in its current form fails to live up to its role as reality's savior. Without Apple's participation and strict social network discipline, AI labeling will remain merely a decorative element. Will we even be able to trust visual information in a couple of years?