you-can-no-longer-rely-on-your-own-eyes-to-determine-what-is-real-instagram-head-raises-ai-issue-for-photos

You can't trust your eyes to tell you what's real anymore: Instagram head raises AI issue for photos

 • 12525 переглядiв

Instagram head Adam Mosseri summarized 2025 by analyzing 20 images dedicated to the new era of "endless synthetic content" that is becoming harder to distinguish from reality, as well as the old, more personal Instagram feed, which he says has been "dead" for many years, UNN reports with reference to The Verge.

Details

Last year, one of The Verge's writers wrote that "...the default assumption is that a photograph is fake because creating realistic and plausible fake photographs has now become trivial," and, as stated, Mosseri agrees.

"For most of my life, I could safely assume that photos or videos were largely accurate depictions of moments that happened. Obviously, that's no longer the case, and it will take us years to adapt," he noted.

"We will move from assuming that what we see is real by default to starting with skepticism. Paying attention to who is sharing something and why. It will be uncomfortable - we are genetically predisposed to believe our eyes," Mosseri pointed out.

According to Mosseri, the evolution needed for Instagram and other platforms is that "we need to create better tools for creativity. To notice AI-generated content and verify its authenticity. To detect authenticity signals in who is posting content. To continue to improve ranking for originality."

Mosseri, who is focused on Instagram, claims that "we love to complain about 'AI junk,' but there's a lot of amazing AI content," without naming anything specific or specifically mentioning Meta's pursuit of AI tools. He argues that camera manufacturers are going down the wrong path by trying to enable everyone to "look like a professional photographer from 2015."

Instead, he says, "raw," unflattering images temporarily serve as an indicator of reality until AI can copy imperfections as well. Then, "we will need to shift our focus from what is being said to who is saying it," using fingerprints and cryptographic signing of images from the cameras that took them to identify genuine media files instead of relying on tags and watermarks added to AI.

Mosseri, the publication writes, is far from the first tech executive to point out the same problem. Samsung executive Patrick Chomet held the view that "there are no truly real photos" after last year's controversy over Galaxy phones' approach to photographing the Moon, and Apple's Craig Federighi told the WSJ that he was "concerned" about the impact of AI editing.

Zuckerberg's Meta acquires AI startup Manus, 'everyone was talking about' - media30.12.25, 12:15 • [views_2994]

Popular
News by theme