According to Forbes, Instagram CEO Adam Mosseri recently stated on Threads that AI-generated content, or “AI slop,” is now so prevalent across social platforms that we’re seeing an “abundance” of it. He predicts that within just a few years, there will be more content created by AI than captured by traditional means like cameras. Mosseri argued that while platforms will initially get better at identifying AI content, they’ll actually get worse at it over time as the technology improves. His proposed solution is a fundamental shift: instead of trying to tag all the fakes, it will be more practical to “fingerprint” and track real media created by humans. This could involve camera makers adding digital watermarking to authentic photos. He also referenced claims, though unverified, that AI content might already account for as much as 70% of what we see online.
The Admission We Saw Coming
Here’s the thing: Mosseri’s post feels less like a revelation and more like a public surrender. And honestly, it’s a relief to hear someone in his position just say it. We’ve all been scrolling, squinting at a photo, and wondering, “Is that real?” The arms of despair have been creeping up for a while now. His point about platforms getting worse at detection over time is the real kicker. It inverts the entire problem. We assumed tech would save us, but it’s the very thing making the situation impossible.
Fingerprinting Reality: A Sensible Hail Mary
So, his idea to fingerprint real content is fascinating. Basically, you start from a position of trust with verifiable human creation and work backwards. The comparison to stock photo metadata is apt, but there’s a massive gap between a curated site like Unsplash and the chaotic firehose of Instagram Stories. Getting every camera maker and phone manufacturer on board to embed inescapable watermarks? That’s a huge technical and commercial coordination challenge. And what about all the existing, legitimate content already out there without a fingerprint? It creates a two-tiered history of media.
The Slop Is Already Winning
Look, the LinkedIn example he indirectly references is proof this is already failing. Automated tagging systems are notoriously brittle. I’ve seen them miss obvious AI images and, probably worse, falsely flag real human work. When the volume is this high, accuracy plummets. Mosseri’s stance is a pragmatic, if depressing, acknowledgment of scale. It’s not about perfection anymore; it’s about creating little islands of verifiable truth in an ocean of generated noise. The goalposts have moved from “identify the fake” to “protect the real.”
What Credit Even Looks Like Now
This all leads to a bigger, weirder question: what does “credit” mean in 2026? If we can technically verify a human creator, does that make their content more valuable? Will feeds prioritize “fingerprinted” posts? There’s a potential future where authenticity becomes a premium feature, a badge you pay for or a tool for professionals. For the average user, though, will they even care? The conversation has shifted from detection to economics and reputation. It’s no longer a game of “spot the bot,” but “who do you trust?” And that’s a much harder game to code for.
