Can Technology Prove the Reliability of Smartphone Photos in the Age of AI-Generated Fakes?

Based on an article by T.J. Thomson, Associate Professor of Visual Communication & Digital Media at RMIT University, reported in The Conversation - Technology.

AI-Generated Fakes Undermine the Credibility

of Video Evidence

The proliferation of generative AI has led to the routine circulation of images that make world leaders appear to make non-existent statements, and videos where celebrities seem to endorse fictional products. Even footage of disaster zones frequently turns out not to depict real events.

We now regularly see a mix of the surreal and the fantastic. World leaders saying things they didn’t. Public figures seemingly endorsing fake products. And disasters that unfold before our eyes but didn’t really happen. Now, anyone can create photorealistic images or videos using AI for a range of purposes. This has changed the nature of visual evidence and our relationship to photos and videos, more broadly.

In an era where anyone can easily generate near-photographic images and videos with AI, we have entered a situation where footage is no longer unconditionally trusted as “evidence of a captured fact.” This growing skepticism applies even to everyday scenes: a photo confirming an UberEats delivery, a video recording a collision on a highway, or an image showing damage to an online purchase. Voices questioning the authenticity of visual media are becoming louder.

Google and Apple Tackle Content Authentication

In response to these challenges, major technology companies are beginning to introduce systems that prove the origin of photos. Google has been embedding content authentication software into the cameras of its own smartphones since 2025. Major camera manufacturers like Nikon, Sony, and Canon have also adopted a similar technical standard, “C2PA,” in their standalone cameras.

Apple is also expected to join this trend. In early August 2026, multiple media outlets reported that Apple’s next operating system would feature a “reference image” function. Similar to the software used in Google smartphones, this technology has the potential to prove that a photograph was genuinely transmitted from the user’s device.

The mechanism works as follows: The raw image and its accompanying metadata are sent to Apple for verification. Upon completion, the image is assigned a unique ID. This ID allows third parties to verify when the photo was taken and on which device.

Challenges and Limitations

However, this technology comes with several constraints.

First, the feature is expected to be disabled by default. Users would need to manually enable it.

Second, it will not be applied retroactively to existing photos. Even after being turned on, it does not automatically apply to all future photos. Before taking a picture, a user must decide “this photo will need its authenticity verified later” and individually enable the setting.

But there are, of course, caveats. The limits of visual evidence. First, the feature, if it is rolled out, will allegedly be turned off by default. Users will have to manually turn it on. Second, once turned on, the feature doesn’t work retroactively with your existing photos. It also doesn’t work automatically with all the future photos you’ll take.

Third, there is the problem that even a “verified” photo can still be intentionally misleading. The distance from the subject, the angle, and a lack of context can give viewers a false impression without actually distorting the facts themselves.

C2PA Adoption and Industry Trends

C2PA (Coalition for Content Provenance and Authenticity) is a technical standard for recording the origin and modification history of content. It is being promoted by an industry consortium that includes Google, Adobe, Microsoft, Intel, Nvidia, and the BBC. Adoption has been expanding since 2025 in both smartphone cameras and standalone cameras.

If Apple introduces its own implementation, compatibility with C2PA will be a key focus. Fragmentation of technical standards could undermine the trustworthiness of the entire verification ecosystem.

As demonstrated by Google’s recent refresh of the Google Photos UI and Gemini video editing features, there is a growing trend toward centralized platform-side management and editing of images and videos. Content authentication is positioned as part of this broader shift.

Editorial Opinion

The introduction of content authentication technology could, in the short term, become a differentiating factor for the digital camera and smartphone markets. A situation where Apple and Google compete with separate implementations could make ensuring compatibility more difficult. It remains to be seen whether C2PA will become established as an industry standard, with a likely conclusion by 2027. In the long term, it could even influence the standards for handling video evidence in legal proceedings. The value of this technology will completely change if courts or insurance companies come to adopt “authenticated images” as valid evidence. A key issue to examine in the future is how well this technology integrates with new digital materials—such as screenshots, screen recordings, or Reality Capture-generated 3D scenes—if authentication only applies to “the act of taking a photo.” There are structural limits to the extent to which technology can restore the credibility of visual media.

References

• “More phones may soon let us prove photos are real – but will it solve the AI fake image crisis?”, by T.J. Thomson, Associate Professor of Visual Communication & Digital Media, RMIT University — The Conversation - Technology, 2026-08-23T20:05:59.000Z (CC BY-ND)
• Source URL: https://theconversation.com/more-phones-may-soon-let-us-prove-photos-are-real-but-will-it-solve-the-ai-fake-image-crisis-290051