It’s 2026, and AI-generated images and videos are virtually indistinguishable from reality, especially to the untrained eye. The question of what is real has moved from a philosophical debate to a daily business challenge. As reported by MIT Technology Review, Microsoft has recently put forward a comprehensive blueprint to help the world prove what’s real online.
The search for answers in a post-truth digital economy
Microsoft is leading a significant new initiative designed to prove the provenance of digital content. By championing a ‘gold standard’ for content verification, Microsoft is attempting to build a digital fingerprint for every image and video created by humans (and robots).
As Microsoft’s Chief Scientific Officer, Eric Horvitz, explains, the goal isn’t for Big Tech to decide what’s true, but to provide the technical labels that tell people where content came from.
The gold standard uses provenance, watermarks, and fingerprints
Imagine you have a Rembrandt painting. To prove it’s an original, you wouldn’t just look at it; you’d provide a detailed manifest of its history (provenance), check for invisible marks (watermarking), and perhaps perform a digital scan of the unique brush strokes (fingerprints).
Microsoft’s AI safety team evaluated 60 different combinations of these methods to find the most resilient shield against modern deception. Their blueprint recommends a three-layered defence:
Cryptographic provenance: A digital manifest that tracks a file from the moment it was captured.
Invisible watermarking: Machine-readable signals that survive even if a file is cropped or screenshotted.
Mathematical signatures: A fingerprint of the content that breaks if even a few pixels are inconsequentially changed.
C2PA metadata
At the heart of Microsoft’s plan is the Coalition for Content Provenance and Authenticity (C2PA) standard. The C2PA provides an open technical standard for publishers, creators and consumers to establish the origin and edits of digital content. It’s called Content Credentials, and it ensures content complies with standards as technology evolves.
Think of this as a digital nutritional label for media. When a content creator takes a photo or films a behind-the-scenes video, invisible metadata is baked into the file at the moment of creation. This metadata tracks the history of the asset, including:
- which device captured it
- whether or not it’s been edited.
- which AI tools (if any) were used to modify it.
Microsoft is now integrating this across its products, from Bing Search and Microsoft 365 to Copilot and Bing Image Creator. When a user sees a piece of content, they can click a small crumb icon to see exactly where it came from and how much of it is original.
Why should content be verified?
To understand why Microsoft is pushing so hard for content verification in 2026, we looked back at a pivotal moment in 2021. Long before generative AI became a household tool, Microsoft recognised that the Photoshop era was evolving into something far more dangerous: deepfakes and disinformation.
As outlined in its foundational 2021 briefing, Microsoft joined forces with Adobe, Intel, and other global technology companies to create C2PA. This was the start of a multi-year mission to build a permanent, unhackable history for every piece of digital media.
Despite these joint efforts, the industry isn’t moving in perfect sync. An audit by Indicator tested automated labelling systems of major social media platforms. The researchers posted a variety of AI-generated content across Instagram, LinkedIn, Pinterest, TikTok, and YouTube to see if the platforms’ internal detection tools would correctly flag the media.
The results showed that only 30% of the test posts were accurately labelled, revealing a significant gap between the platforms’ public promises of transparency and the actual performance of their AI detection systems.
The report further noted that this failure isn’t just technical. Leaders like Mark Zuckerberg or Elon Musk face a conflict of interest: if they believe AI-generated labels will cause users to stop scrolling or reduce engagement, they have a financial incentive to be less aggressive with their labelling enforcement.
Ultimately, Microsoft’s initiative is a reminder that in a world of infinite AI, the truth is a luxury.
